From kaffe@kaffe.org  Wed Oct  1 14:42:59 1997
From: kaffe@kaffe.org (Mikael Ståldal)
Date: Wed, 01 Oct 1997 14:42:59 +0100
Subject: Porting Kaffe to OS/2
Message-ID: <jNlM0EpfhKYU092yn@ingen.reklam.nada.kth.se>

I'm trying to port Kaffe to OS/2 (an i386 platform) using the EMX compiler
(GCC port). I have encountered some problems.

Some different header files differ only in case, and since the file system
in OS/2 isn't case sensitive this causes clashes. String.h clashes with the
standard string.h and System.h clashes with Kaffe's system.h.

Make doesn't find machine.c in the interp subdirectory, how is it supposed
to find it there? I'm using GNU Make 3.75. When I copied it to kaffevm it
works. The same problem with the package directoris in packages.

When I copied files from the packages, things broke due to "../../.." style
#include directives that I hade to change manually. Try to avoid using ".."
in #include directives.

I get the following error:

make[1]: Entering directory `/UNIX/kaffe/kaffe/kaffevm'
gcc -g -O -Zexe -DNO_SHARED_LIBRARIES -I. -I. -I./../../config -I../../config -I../../include -I./../../include -DINTERPRETER -I./intrp   -DKVER=\"0.91\" -c  machine.c
machine.c: In function `virtualMachine':
machine.c:141: structure has no member named `exception_table_len'
machine.c:152: structure has no member named `ins'
machine.c:178: structure has no member named `code'
machine.c:182: structure has no member named `codelen'
make[1]: *** [machine.o] Error 1
make[1]: Leaving directory `/UNIX/kaffe/kaffe/kaffevm'
make: *** [all] Error 1

When I look at the structure, it has no exception_table_len member. How is
this supposed to works?

-- 
/****************************************************************\
* You have just read a message from Mikael Ståldal.              *
*                                                                *
* Remove "ingen.reklam." from the address before mail replying.  *
\****************************************************************/

From kaffe@kaffe.org  Thu Oct  2 01:06:44 1997
From: kaffe@kaffe.org (Ram Madduluri)
Date: Wed, 01 Oct 1997 18:06:44 -0600
Subject: Kaffe runtime question
Message-ID: <s4328dbf.005@novell.com>

Hello,

I have ported Kaffe 0.9.1 to OSF/1 on x86 platform. The initial
HelloWorldApp test passed. However, when I run a sample server application -
UDPEchoServer, I get the following error:
	$ kaffe UDPEchoServer
	My IP: 127.0.0.1
	My port: 6666
	Failed to locate native function:
        	java/net/PlainDatagramSocketImpl.datagramSocketCreate()V
	java.lang.UnsatisfiedLinkError
        	at java/net/DatagramSocket.create(133)
        	at java/net/DatagramSocket.<init>(114)
        	at java/net/DatagramSocket.<init>(100)
        	at UDPServer.main(41)

The PlainDatagramSocketImpl class is in the
/usr/local/share/kaffe/classes.zip file. Am I missing something here? I can
run some of the other tests just fine (DemoServer, DemoClient, ..). Is there
some configuration stuff that I messed up??

Any pointers would be greatly appreciated.

Regards,

Ram
--
Ram Madduluri                                                  E-Mail:
ram@Novell.COM
Corporate Research                                          Voice:  
801-222-5361
Novell Inc.                                                          Fax:   
  801-222-5177
1555 North Technology Way
Orem, UT 84097

From kaffe@kaffe.org  Thu Oct  2 05:33:54 1997
From: kaffe@kaffe.org (Orville R. Weyrich)
Date: Wed, 1 Oct 1997 21:33:54 -0700 (MST)
Subject: Porting Kaffe to OS/2
In-Reply-To: <jNlM0EpfhKYU092yn@ingen.reklam.nada.kth.se>
Message-ID: <Pine.LNX.3.91.971001210131.6624A-100000@dopey.weyrich.com>

On Wed, 1 Oct 1997, Mikael Ståldal wrote:

> I'm trying to port Kaffe to OS/2 (an i386 platform) using the EMX compiler
> (GCC port). I have encountered some problems.
> 
> Some different header files differ only in case, and since the file system
> in OS/2 isn't case sensitive this causes clashes. String.h clashes with the
> standard string.h and System.h clashes with Kaffe's system.h.

My suggestion is to place all kaffe include files in a directory named
kaffe, such that the kaffe directory is in your INCLUDE path.  This means
that you will have to modify the source code to:

	#include <string.h>

and

	#include <kaffe/String.h>

It is less work if you only do this for the files which conflict, but it
is cleaner if you move all the kaffe headers to this location and tweak the
source code appropriately.  In my opinion, this would add value to the 
source code base for all users, since it would make clear which includes
are for system headers, and which are for kaffe headers.

I just downloaded kaffe myself, and haven't looked at the code yet,
so some other variation may be more consistent with the existing
organization, but this should get you started.  I use OS/2 and would
love to have kaffe on OS/2, so give me a shout if I can be of any
more help :-)

> 
> Make doesn't find machine.c in the interp subdirectory, how is it supposed
> to find it there? I'm using GNU Make 3.75. When I copied it to kaffevm it
> works. The same problem with the package directoris in packages.

Is there a configuration script that generates machine.c based on the
environment?
 
> When I copied files from the packages, things broke due to "../../.." style
> #include directives that I hade to change manually. Try to avoid using ".."
> in #include directives.

I am afraid that there is no general solution for this portability problem,
absolute paths also have their problems.  Perhaps a perl script containing
something like

	if ( $line =~ /^#include/ )
	{
		$line =~ s/\.\.\//..\\/;
	}

might be the easiest to do over all your source files.
(i.e. find all include lines and change ../ to ..\).

> 
> I get the following error:
> 
> make[1]: Entering directory `/UNIX/kaffe/kaffe/kaffevm'
> gcc -g -O -Zexe -DNO_SHARED_LIBRARIES -I. -I. -I./../../config 
-I../../config -I../../include -I./../../include -DINTERPRETER -I./intrp   -DKVER=\"0.91\" -c  machine.c
> machine.c: In function `virtualMachine':
> machine.c:141: structure has no member named `exception_table_len'
> machine.c:152: structure has no member named `ins'
> machine.c:178: structure has no member named `code'
> machine.c:182: structure has no member named `codelen'
> make[1]: *** [machine.o] Error 1
> make[1]: Leaving directory `/UNIX/kaffe/kaffe/kaffevm'
> make: *** [all] Error 1
> 
> When I look at the structure, it has no exception_table_len member. How is
> this supposed to works?
> 

I don't know.  It may be due to using the wrong header file.

===================================================================
Orville R. Weyrich, Jr.                 Weyrich Computer Consulting
mailto:orville@weyrich.com                   http://www.weyrich.com
-------------------------------------------------------------------

From kaffe@kaffe.org  Thu Oct  2 13:14:41 1997
From: kaffe@kaffe.org (David Turner)
Date: Thu, 2 Oct 1997 13:14:41 +0100 (MET DST)
Subject: Porting Kaffe to OS/2
In-Reply-To: <jNlM0EpfhKYU092yn@ingen.reklam.nada.kth.se> from "Mikael Ståldal" at Oct 1, 97 02:42:59 pm
Message-ID: <199710021114.NAA08384@messiaen.enst.fr>

Hi Mikael

Mikael Ståldal said:
> 
> machine.c: In function `virtualMachine':
> machine.c:141: structure has no member named `exception_table_len'
> machine.c:152: structure has no member named `ins'
> machine.c:178: structure has no member named `code'
> machine.c:182: structure has no member named `codelen'
> make[1]: *** [machine.o] Error 1
> make[1]: Leaving directory `/UNIX/kaffe/kaffe/kaffevm'
> make: *** [all] Error 1
> 
> When I look at the structure, it has no exception_table_len member. How is
> this supposed to works?
>

It's a famous bug in the 0.9.1 release (it's not EMX specific ;-)
The fix has been released in the mail archive, but I'll give it here
again, from memory :

  1. replace 'method->codelen' by 'method->c.bcode.codelen'  (line 182)
     and     'method->code'    by 'method->c.bcode.code'     (line 178)

     ( I don't remember if it's "method" or "meth" or "m", but it's
       a pointer to a method structure )

  2. replace 'exception_table_len > 0' by 'exception_table != 0' (line 141)

  3. define a local variable "char* str" and replace
     line 152 with :

      "str = meth->signature->data;
       idx = sizeofSig( &str, false ) + ( .... & ..... ? ... : ... )'
                                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                                          same as before..

  Then recompile, it should work OK now.. ( well, there are still some
  bugs in there ;-)

Good luck, you'll need it..

- David

From kaffe@kaffe.org  Thu Oct  2 17:59:03 1997
From: kaffe@kaffe.org (Tony Kimball)
Date: Thu, 2 Oct 1997 11:59:03 -0500 (CDT)
Subject: Porting Kaffe to OS/2
References: <jNlM0EpfhKYU092yn@ingen.reklam.nada.kth.se>
 <Pine.LNX.3.91.971001210131.6624A-100000@dopey.weyrich.com>
Message-ID: <199710021659.LAA06487@compound.east.sun.com>

Quoth Orville R. Weyrich on Wed, 1 October:
:  
: > When I copied files from the packages, things broke due to "../../.." style
: > #include directives that I hade to change manually. Try to avoid using ".."
: > in #include directives.
: 
: I am afraid that there is no general solution for this portability problem,
: absolute paths also have their problems. 

-I fixes it everywhere.

From kaffe@kaffe.org  Thu Oct  2 18:49:10 1997
From: kaffe@kaffe.org (James A. Hillyerd)
Date: Thu, 2 Oct 1997 10:49:10 -0700 (PDT)
Subject: Next Release?
Message-ID: <Pine.LNX.3.95.971002104801.6765B-100000@animal.blarg.net>

Any guesses on when the next release of Kaffe is coming out?  This list
has been pretty quiet lately. =)

-james

[  James A. Hillyerd (JH2162) - james@blarg.net - Web Developer  ]
[  http://www.blarg.net/~james/   http://www.hyperglyphics.com/  ]
[ 1024/B11C3751 CA 1C B3 A9 07 2F 57 C9  91 F4 73 F2 19 A4 C5 88 ]

From kaffe@kaffe.org  Sat Oct  4 04:49:11 1997
From: kaffe@kaffe.org (Michael Hirsch)
Date: Fri, 3 Oct 1997 23:49:11 -0400 (EDT)
Subject: java.lang.ClassNotFoundException:HelloWorldApp
Message-ID: <199710040349.XAA22083@cssun.mathcs.emory.edu>

I have the same problem that Bryan Felix reported.

I tried to follow the README instructions exactly.  I got all four
tarballs: 
kaffe-0.9.1.tgz 
kaffe-0.9.1-package-javasoft.com.tgz
kaffe-0.9.1-package-biss-net.com.tgz and
kaffe-0.9.1-package-epfl.ch.tgz 
and unpacked them in the same directory.

./configure worked fine, as did make and make install.

But now:
[34] test> echo $CLASSPATH $KAFFEHOME
.:/usr/local/share/kaffe/classes.zip:/usr/local/share/kaffe/biss.zip /usr/local/share/kaffe
[35] test> pwd
/usr/local/Uploads/kaffe-0.9.1/test
[36] test> kaffe HelloWorldApp
java.lang.ClassNotFoundException: HelloWorldApp
[37] test> 

I tried to do exactly as it says.  . is in the CLASSPATH.
I did not compile HelloWorldApp.fava into a .class file because 1) I
don't know how, exactly, and 2) the README doesn't say to.

However, if I try
[37] test> javac HelloWorldApp.java
my screen fills with this error message:
java.lang.NullPointerException
	at java/io/File.list(line unknown, pc 0x8520a6c)
	at sun/tools/java/ClassPathEntry.getFiles(line unknown, pc 0x8532197)
	at sun/tools/java/ClassPath.getFile(line unknown, pc 0x852ddb9)
	at sun/tools/java/ClassPath.getFile(line unknown, pc 0x8530bd6)
	at sun/tools/java/Package.getBinaryFile(line unknown, pc 0x853477a)
	at sun/tools/java/Package.classExists(line unknown, pc 0x852083b)
	at sun/tools/javac/BatchEnvironment.classExists(line unknown, pc 0x852c650)
	at sun/tools/java/Environment.classExists(line unknown, pc 0x8530369)
	at sun/tools/java/Environment.classExists(line unknown, pc 0x8530369)
	at sun/tools/java/Imports.importable(line unknown, pc 0x851ee55)
	at sun/tools/java/Imports.resolve(line unknown, pc 0x852d2db)
	at sun/tools/java/ImportEnvironment.resolve(line unknown, pc 0x8521eb3)
	at sun/tools/java/Environment.resolve(line unknown, pc 0x85205a2)
	at sun/tools/java/Environment.resolveName(line unknown, pc 0x851dea4)
	at sun/tools/java/Environment.resolveName(line unknown, pc 0x851dcca)
	at sun/tools/javac/SourceClass.resolveSuper(line unknown, pc 0x851ecd4)
	at sun/tools/javac/SourceClass.resolveSupers(line unknown, pc 0x852e895)
	at sun/tools/javac/SourceClass.resolveTypeStructure(line unknown, pc 0x852e1bd)
	at sun/tools/javac/SourceClass.basicCheck(line unknown, pc 0x851f971)
	at sun/tools/java/ClassDeclaration.getClassDefinition(line unknown, pc 0x851eb11)
	at sun/tools/javac/Main.compile(line unknown, pc 0x81b9992)
	at sun/tools/javac/Main.main(line unknown, pc 0x81a1d83)
error: An exception has occurred in the compiler; please file a bug report (http://java.sun.com/cgi-bin/bugreport.cgi).
1 error

so clearly something is wrong with my setup.  I'm sure I'm missing
something obvious.

I tried 'make test', but each test failed with what appears to be
essentially the same error as the one quoted above.

Thanks in advance,

-- 
Michael D, Hirsch			Work: (404) 727-7940 
Emory University, Atlanta, GA 30322	FAX: (404) 727-5611
email:  hirsch@mathcs.emory.edu		http://www.mathcs.emory.edu/~hirsch/

Public key for encrypted mail available upon request (or finger
hirsch@cssun.mathcs.emory.edu).

From kaffe@kaffe.org  Sun Oct  5 06:57:09 1997
From: kaffe@kaffe.org (Henrik Martin)
Date: Sat, 4 Oct 1997 22:57:09 -0700 (PDT)
Subject: Working version of kaffe for BSDI 3.0?
Message-ID: <199710050557.WAA29595@u2.netgate.net>


Hi folks! I'm trying to build kaffe v0.9.1 on a x86 machine running
BSDI 3.0, but it won't build. After having changed the "global" to "globl"
in the config/i386/trampolines.c file, I get the following linking error:

make -k
rm -f libkaffe_vm.a
ln -s libkaffe_vm.a.0.91 libkaffe_vm.a
gcc -g -O  -I. -I./../kaffevm -I../../config -I./../../config -I../../include -I./../../include -o kaffe main.o -L../kaffevm -lkaffe_vm  -L../../packages/tjwassoc.co.uk/APIcore/lib -lkaffe_native -L../../packages/tjwassoc.co.uk/APInet/lib -lkaffe_net -L../../packages/tjwassoc.co.uk/APIzip/lib -lkaffe_zip  -lipc -L/usr/X11R6/lib -lX11 -lkaffe_vm -lipc -lm -lc
ld: premature end of file in symbols of ../kaffevm/libkaffe_vm.a(machine.o)
*** Error code 1 (continuing)
`all' not remade because of errors.

I do have the kaffe-0.9.1-package-javasoft.com.tgz stuff installed
BTW.  Any hints on what may be wrong? If anyone has built any recent
version of kaffe successfully on BSDI 3.0, please let me know.

If you reply to the mailing list, please also cc: henrik@netgate.net.

Thanks,

/Henrik

From kaffe-announce@kaffe.org  Sun Oct  5 15:22:07 1997
From: kaffe-announce@kaffe.org (Tim Wilkinson)
Date: Sun, 05 Oct 1997 15:22:07 +0100
Subject: [FINALLY] Kaffe 0.9.2 released.
Message-ID: <3437A28F.41C67EA6@tjwassoc.co.uk>

KAFFE v0.9.2 - A JIT and interpreting virtual machine to run Java(tm)*
code
===========================================================================

                *** Kaffe is now JDK 1.1.2 compliant ***

Kaffe is a virtual machine design to execute Java bytecode.
This machine can be configured in two modes.  In one mode it operates as
a pure bytecode interpreter (not unlike Javasoft's machine); in the
second
mode if performs "just-in-time" code conversion from the abstract code
to
the host machine's native code.  This will ultimately allow execution of
Java code at the same speed as standard compiled code but while
maintaining
the advantages and flexibility of code independence.

New Features
------------

This version of Kaffe has the following new features:

 * Kaffe is now JDK 1.1.3 compliant !

 * Improved memory efficiency when allocating object.

 * Improvements to GC.
 
 * Alpha port partially fixed ... still lots of problems though. 
 
 * Lots of bug fixes (see Changelog)

Platforms
---------

Kaffe will run on the following platforms ('J' indicates 'just-in-time'
support, 'I' indicates 'interpreter' support):

 
========================================================================
  uProc     | Mode |    Systems
 
========================================================================
  i386      | J+I  |    FreeBSD 2.x +   Unixware        NetBSD 1.x
            |      |    Solaris 2.x     BSDI 2.x        Linux 2.0.0 +
            |      |    SCO 3.2v5       NeXTStep 3.x    Windows '95
            |      |    DG/UX           OpenBSD 2.x     OpenStep 4.x
            |      |    QNX 4.x
 
------------------------------------------------------------------------
  Sparc     | J+I  |    SunOS 4.x       Solaris 2.x     NetBSD 1.x
            |      |    NeXTStep 3.x    Linux           Fujitsu UXP/DS
            |      |    BSDI 3.x
 
------------------------------------------------------------------------
  Alpha     | J+I  |    Linux           OSF/1 
 
------------------------------------------------------------------------
  M68K      | J+I  |    AmigaOS         NeXTStep 3.x    NetBSD 1.x
            |      |    SunOS 4.x       Linux           AUX
            |      |    OpenBSD 2.x
 
------------------------------------------------------------------------
  PowerPC   | I    |    MkLinux         MachTen 4.0.3   AIX
 
------------------------------------------------------------------------
  MIPS      | I    |    IRIX 5 & 6      NetBSD 1.x
 
------------------------------------------------------------------------
  PARISC    | I    |    HPUX 10.x
 
------------------------------------------------------------------------  
  StrongARM | I    |    RiscIX          (work-in-progress)
 
========================================================================

I encourage people to make ports to other platforms and to feed them
back
to me.  I would recommend that initial ports concentrate on the
interpreter.
These can then be used as the basis for developing JIT versions for
these
systems.

The sources for Kaffe can be found at the following location:

Europe:
        ftp://ftp.kaffe.org/pub/kaffe/
        ftp://ftp.sarc.city.ac.uk/pub/kaffe/
        ftp://ftp.doc.ic.ac.uk/packages/kaffe/
        ftp://ftp.lysator.liu.se/pub/languages/java/kaffe/
        ftp://sunsite.mff.cuni.cz/Languages/Java/kaffe/
        ftp://sunsite.auc.dk/pub/languages/java/kaffe/
        ftp://alterego.stack.urc.tue.nl/pub/kaffe/
        ftp://ftp.fh-wolfenbuettel.de/pub/lang/java/kaffe/
        ftp://ftp.uni-erlangen.de/pub/source/Java/kaffe/
        ftp://gd.tuwien.ac.at/languages/java/kaffe/
USA:
        ftp://ftp.cs.columbia.edu/pub/kaffe/
        ftp://sunsite.unc.edu/pub/languages/java/kaffe/
Japan:
        ftp://ftp.webcity.co.jp/pub/andoh/java/kaffe/
Others:
        ftp://ftp.unicamp.br/pub/languages/java/kaffe/


What you need
-------------

For a Kaffe system you need the following parts:

 kaffe-0.9.2.tgz                        Virtual machine and all bits
                                        provided by tjwassoc.co.uk
 kaffe-0.9.2-package-javasoft.com.tgz   Sun's class libraries.

These two are are the minimum to get something useful working (you might
already have the classes libraries from JDK 1.1.2; if so don't bother to
get them again).

If you want to use graphics you'll need one of the following:

 kaffe-0.9.2-package-biss-net.com.tgz   AWT graphics support.

Uncompress and untar these archives into the same directory, configure,
make and install.

Enjoy.

Tim Wilkinson
<tim@kaffe.org>
----
* Java is a registered trademark of Sun Microsystems, Inc.

--
  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
  London, UK.                           Email:   tim@tjwassoc.co.uk

From kaffe@kaffe.org  Sun Oct  5 17:00:14 1997
From: kaffe@kaffe.org (Matthias Hopf)
Date: Sun, 5 Oct 1997 18:00:14 +0200 (MET DST)
Subject: Next Release?
In-Reply-To: <Pine.LNX.3.95.971002104801.6765B-100000@animal.blarg.net> from "James A. Hillyerd" at Oct 2, 97 10:49:10 am
Message-ID: <199710051600.SAA26186@faui01.informatik.uni-erlangen.de>

> Any guesses on when the next release of Kaffe is coming out?  This list
> has been pretty quiet lately. =)
> 
> -james

Kaffe v0.9.2 has just been released. But as I have to learn for final
exams right now, I won't be able to do anything about kaffe until
end of october. So expect a tested version of V0.9.2 (or V0.9.3 then ;)
in november.

And yes, it has been quiet here the last months. Nothing happened about
kaffe either, lots of the developers seemed to be on holiday ;)
I didn't do anything about kaffe0.9.1 as it seemed to have several
unfixed bugs.

CU

Matthias

-- 
    //             |   Matthias Hopf - "Hoeppel"   |    _      __
\\ //    Amiga     |  student of computer science  |  _|cience |-iction
 \X/ by conviction |      in Erlangen/Germany      | by belief in Future

 EMail:  mshopf@informatik.uni-erlangen.de
 Aminet: ftpamiga@epix.rrze.uni-erlangen.de
 WWW:    http://wwwcip.informatik.uni-erlangen.de/user/mshopf/

From kaffe@kaffe.org  Sun Oct  5 17:59:23 1997
From: kaffe@kaffe.org (Mark Huizer)
Date: Sun, 5 Oct 1997 18:59:23 +0200
Subject: [FINALLY] Kaffe 0.9.2 released.
In-Reply-To: <3437A28F.41C67EA6@tjwassoc.co.uk>; from "Tim Wilkinson" on Sun, Oct 05, 1997 at 03:22:07PM +0100
References: <3437A28F.41C67EA6@tjwassoc.co.uk>
Message-ID: <19971005185923.36519@xaa.stack.nl>

> 
>                 *** Kaffe is now JDK 1.1.2 compliant ***
> 
>  * Kaffe is now JDK 1.1.3 compliant !

Nice :)

What's the status of the BISS AWT classes by now? I missed out a few
weeks, so there's the lack of knowledge. Is it possible to run graphical
stuff a bit better by now?
-------------------------------------------------------------------------
- Mark Huizer  - xaa@stack.nl     - rcbamh@urc.tue.nl                   -
-------------------------------------------------------------------------
- Anyone who is capable of getting themselves made President should	-
- on no account be allowed to do the job (The Hitchhikers Guide)	-
-------------------------------------------------------------------------

From kaffe@kaffe.org  Sun Oct  5 19:53:36 1997
From: kaffe@kaffe.org (Bryan Felix)
Date: Sun, 5 Oct 1997 11:53:36 -0700 (MST)
Subject: Working version of kaffe for BSDI 3.0?
Message-ID: <199710051853.LAA20460@rapidaxcess.com>

>Hi folks! I'm trying to build kaffe v0.9.1 on a x86 machine running
>BSDI 3.0, but it won't build. After having changed the "global" to "globl"
>in the config/i386/trampolines.c file, I get the following linking error:
>
>If you reply to the mailing list, please also cc: henrik@netgate.net.
>
>Thanks,
>
>/Henrik
>
        I have done my best to get any version running on BSDI 3.0 and on
BSDI 2.1 and finally cut it off. Only so much time to spare on one project.
I decided this project is just to buggy and am waiting patiently for a
stable developement to be released.
        I did manage to get version 0.9.0 to the point that it would at
least just hang after I went through typing in the CLASSPATH by hand with
each execution. I was in touch with another user on this list that has
gotten an even older version to run on BSDI though he has to add the
CLASSPATH on to every execution as well.
        Sorry I can't be of much more help, but thought I would add my
experience for what it's worth.

With Regret,
Bryan Felix


                            _______________________________________________
        $$$$$$$$$$$$$$$$               ---Rapid Axcess---
          $$$$$$$$$$$$$$$$$    ...Premium Web Design & Domain Hosting...
           $$$$     $$$$$$$$     ______________________________________ 
           $$$$      $$$$$$$    (_____       Bryan Felix        _______)
           $$$      $$$$$   __________) felix@rapidaxcess.com  (__
          $$$     $$$$$    (_______      Voice: 602-952-8275  ______)
         $$$$  $$$$$           ____)     Page: 602-201-4646  (____________
        $$$$$$$$$             (__________ Fax: 602-224-9585 ______________)
       $$$   $$$$        $$$        
      $$$     $$$$      $$ $$       PO Box 61413 
    $$$        $$$$$   $$___$$      Phoenix, AZ 85082-1413
 $$$$$$        $$$$$$$$$     $$ ____http://www.rapidaxcess.com_____________

From kaffe@kaffe.org  Sun Oct  5 22:32:10 1997
From: kaffe@kaffe.org (Michael Hirsch)
Date: Sun, 5 Oct 1997 17:32:10 -0400 (EDT)
Subject: java.lang.ClassNotFoundException:HelloWorldApp
In-Reply-To: <199710040349.XAA22083@cssun.mathcs.emory.edu>
References: <199710040349.XAA22083@cssun.mathcs.emory.edu>
Message-ID: <199710052132.RAA23133@cssun.mathcs.emory.edu>

Okay, I found some of the problem.  I wrote:

>>>>> In article <199710040349.XAA22083@cssun.mathcs.emory.edu>, Michael Hirsch <hirsch@mathcs.emory.edu> writes:

Michael> I have the same problem that Bryan Felix reported.  I tried
Michael> to follow the README instructions exactly.  I got all four
Michael> tarballs: kaffe-0.9.1.tgz
Michael> kaffe-0.9.1-package-javasoft.com.tgz
Michael> kaffe-0.9.1-package-biss-net.com.tgz and
Michael> kaffe-0.9.1-package-epfl.ch.tgz and unpacked them in the same
Michael> directory.

Michael> ./configure worked fine, as did make and make install.

Michael> But now: 
Michael> [34] test> echo $CLASSPATH $KAFFEHOME
Michael>
.:/usr/local/share/kaffe/classes.zip:/usr/local/share/kaffe/biss.zip /usr/local/share/kaffe 
MIchael> [35] test> pwd
Michael> /usr/local/Uploads/kaffe-0.9.1/test 
Michael> [36] test> kaffe HelloWorldApp 
Michael> java.lang.ClassNotFoundException: HelloWorldApp

This, of course, does not work because the .java file has not been
compiled into a .class file.  Unfortunately, the README file says that
this should work.  Perhaps the next release will add the compiled
HelloWorldApp.class file to the test directory so silly people like me
don't get confused.

I made a compiled the .java file on another system and then tried to
run it and it worked fine.  So the interpreter/JIT seems to be okay.

Next, I tried compiling and that is still broken.  I reinstalled
everything and I still get these errors.  Any further suggestions?
It kind of looks like I don't have something installed, but I
installed all 4 .tgz files.

Thanks,

--Michael

Here are the errors:
[2] test> javac HelloWorldApp.java 
java.lang.NullPointerException
	at java/io/File.list(line unknown, pc 0x8520a6c)
	at sun/tools/java/ClassPathEntry.getFiles(line unknown, pc 0x8532197)
	at sun/tools/java/ClassPath.getFile(line unknown, pc 0x852ddb9)
	at sun/tools/java/ClassPath.getFile(line unknown, pc 0x8530bd6)
	at sun/tools/java/Package.getBinaryFile(line unknown, pc 0x853477a)
	at sun/tools/java/Package.classExists(line unknown, pc 0x852083b)
	at sun/tools/javac/BatchEnvironment.classExists(line unknown, pc 0x852c650)
	at sun/tools/java/Environment.classExists(line unknown, pc 0x8530369)
	at sun/tools/java/Environment.classExists(line unknown, pc 0x8530369)
	at sun/tools/java/Imports.importable(line unknown, pc 0x851ee55)
	at sun/tools/java/Imports.resolve(line unknown, pc 0x852d2db)
	at sun/tools/java/ImportEnvironment.resolve(line unknown, pc 0x8521eb3)
	at sun/tools/java/Environment.resolve(line unknown, pc 0x85205a2)
	at sun/tools/java/Environment.resolveName(line unknown, pc 0x851dea4)
	at sun/tools/java/Environment.resolveName(line unknown, pc 0x851dcca)
	at sun/tools/javac/SourceClass.resolveSuper(line unknown, pc 0x851ecd4)
	at sun/tools/javac/SourceClass.resolveSupers(line unknown, pc 0x852e895)
	at sun/tools/javac/SourceClass.resolveTypeStructure(line unknown, pc 0x852e1bd)
	at sun/tools/javac/SourceClass.basicCheck(line unknown, pc 0x851f971)
	at sun/tools/java/ClassDeclaration.getClassDefinition(line unknown, pc 0x851eb11)
	at sun/tools/javac/Main.compile(line unknown, pc 0x81b9992)
	at sun/tools/javac/Main.main(line unknown, pc 0x81a1d83)
error: An exception has occurred in the compiler; please file a bug report (http://java.sun.com/cgi-bin/bugreport.cgi).
1 error
[3] test> 

From kaffe@kaffe.org  Mon Oct  6 18:21:09 1997
From: kaffe@kaffe.org (P. V. Thatayya Naidu)
Date: Mon, 6 Oct 1997 10:21:09 -0700
Subject: Object Seralistion!
Message-ID: <199710061721.KAA01323@faith>

Hi,
    Will  Object Serialistion and RMI work with this new release Kaffe 0.9.2 ?
    
    
thanks
naidu
P. V. Thatayya Naidu

From kaffe@kaffe.org  Mon Oct  6 14:03:55 1997
From: kaffe@kaffe.org (Tim Wilkinson)
Date: Mon, 06 Oct 1997 14:03:55 +0100
Subject: BISS-AWT and Kaffe 0.9.2
Message-ID: <3438E1BB.1CFBAE39@tjwassoc.co.uk>

All,

Please note that I've withdrawn the BISS-AWT and EPFH.CH packages from
the latest release since neither of them work at the moment.  Hopefull
BISS-AWT will be back soon.

Regards
Tim
--
  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
  London, UK.                           Email:   tim@tjwassoc.co.uk

From kaffe@kaffe.org  Mon Oct  6 14:45:46 1997
From: kaffe@kaffe.org (Peter C. Mehlitz)
Date: Mon, 6 Oct 1997 15:45:46 +0200
Subject: [FINALLY] Kaffe 0.9.2 released.- BISS-AWT status
In-Reply-To: Mark Huizer <xaa@stack.nl>
 "Re: [FINALLY] Kaffe 0.9.2 released." (Oct  5,  6:59pm)
References: <3437A28F.41C67EA6@tjwassoc.co.uk>
 <19971005185923.36519@xaa.stack.nl>
Message-ID: <9710061545.ZM280@mwave.biss-net.com>

> Is it possible to run graphical stuff a bit better by now?

Sorry, not yet released. We have the biss.jde up ad running again, but before
we go public, this time we want to improve java.awt compatibility. This is even
more important since the java.awt classes are replaced as well. We also still
need the go from our sponsor.

Peter

From kaffe@kaffe.org  Mon Oct  6 17:25:54 1997
From: kaffe@kaffe.org (TIOOYK)
Date: Mon, 6 Oct 1997 09:25:54 -0700
Subject: Working version of kaffe for BSDI 3.0?
References: <199710051853.LAA20460@rapidaxcess.com>
Message-ID: <199710061625.JAA00292@van-as-05b01.direct.ca>

On Sun, Oct 05, 1997 at 11:53:36AM -0700, Bryan Felix wrote:
: / >Hi folks! I'm trying to build kaffe v0.9.1 on a x86 machine running
: / >BSDI 3.0, but it won't build. After having changed the "global" to "globl"
: / >in the config/i386/trampolines.c file, I get the following linking error:
: / >
: / >If you reply to the mailing list, please also cc: henrik@netgate.net.
: / >
: / >Thanks,
: / >
: / >/Henrik
: / >
: /         I have done my best to get any version running on BSDI 3.0 and on
: / BSDI 2.1 and finally cut it off. Only so much time to spare on one project.
: / I decided this project is just to buggy and am waiting patiently for a
: / stable developement to be released.
: /         I did manage to get version 0.9.0 to the point that it would at
: / least just hang after I went through typing in the CLASSPATH by hand with
: / each execution. I was in touch with another user on this list that has
: / gotten an even older version to run on BSDI though he has to add the
: / CLASSPATH on to every execution as well.
: /         Sorry I can't be of much more help, but thought I would add my
: / experience for what it's worth.
: / 

What's the problem? I use BSD/OS 2.1 and running kaffe -version I get:

# kaffe -version
Kaffe Virtual Machine
Copyright (c) 1997 T. J. Wilkinson & Associates, London, UK.
Engine: Just-in-time   Version: 0.91   Java Version: 1.1.2

I don't recall doing much to get things running, apart from the .globl
thing.

Mind you, there is so much that DOESN'T work past the tests that it's
not worth compiling right now : <

To be fair it's the graphic stuff that is pretty broken..  Perhaps
0.9.2 has a useable companion AWT library now??  Guess I'll find out
soon enough.

-g

--
Berbeee-a seys,
 Teke-a qooeeloodes in geen und gu tu a deescu reeght evey!
 Bork Bork Bork!
 Boot Kee seys, VOO-VOO!!  Nu credeet et ``Mr. Leeqoour''!!

From kaffe@kaffe.org  Mon Oct  6 23:34:36 1997
From: kaffe@kaffe.org (Tim Wilkinson)
Date: Mon, 06 Oct 1997 23:34:36 +0100
Subject: Object Seralistion!
References: <199710061721.KAA01323@faith>
Message-ID: <3439677C.3F54BC7E@tjwassoc.co.uk>

P. V. Thatayya Naidu wrote:
> 
> Hi,
>     Will  Object Serialistion and RMI work with this new release Kaffe 0.9.2 ?

I'm afraid not.  Any takers?

Cheers
Tim
--
  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
  London, UK.                           Email:   tim@tjwassoc.co.uk

From kaffe@kaffe.org  Tue Oct  7 03:22:13 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Tue, 07 Oct 1997 11:22:13 +0900
Subject: Working version of kaffe for BSDI 3.0?
In-Reply-To: Your message of Mon, 06 Oct 1997 09:25:54 MST.
Message-ID: <18855.876190933@rananim.ie.u-ryukyu.ac.jp>

Hi.

>What's the problem? I use BSD/OS 2.1 and running kaffe -version I get:

In BSD/OS 2.1 and 0.8.4, Kaffe works, but after 0.9.0, they use
stack offset in setjmp buf, which is not correnct in BSD/OS 3.0.
Ya, Bryan Felix was correct. It didn't work at all.

Here is a fix for 0.9.2

In config/i386/bsdi/md.h,  add SP_OFFSET redefine.

#include "i386/common.h"
#include "i386/threads.h"

#undef SP_OFFSET
#define SP_OFFSET               2

-- 
Shinji KONO @ Information Engineering, University of the Ryukyus
$B2OLn??<#(B @ $BN05eBg3X9)3XIt>pJs9)3X2J(B

From kaffe@kaffe.org  Tue Oct  7 07:45:29 1997
From: kaffe@kaffe.org (Orville R. Weyrich)
Date: Mon, 6 Oct 1997 23:45:29 -0700 (MST)
Subject: [FINALLY] Kaffe 0.9.2 released.
In-Reply-To: <3437A28F.41C67EA6@tjwassoc.co.uk>
Message-ID: <Pine.LNX.3.91.971006234218.28957C-100000@dopey.weyrich.com>

On Sun, 5 Oct 1997, Tim Wilkinson wrote:

>                 *** Kaffe is now JDK 1.1.2 compliant ***
 
> This version of Kaffe has the following new features:
> 
>  * Kaffe is now JDK 1.1.3 compliant !

Is one of these a typo, or is it compliant with both?

> What you need
> -------------

> 
> These two are are the minimum to get something useful working (you might
> already have the classes libraries from JDK 1.1.2; if so don't bother to
> get them again).

I have JDK 1.1.3 -- do I need to get 1.1.2, or will either work?

Thanks

orville

===================================================================
Orville R. Weyrich, Jr.                 Weyrich Computer Consulting
mailto:orville@weyrich.com                   http://www.weyrich.com
-------------------------------------------------------------------

From kaffe@kaffe.org  Tue Oct  7 13:48:38 1997
From: kaffe@kaffe.org (Gregor Hoffleit)
Date: Tue,  7 Oct 97 14:48:38 +0200
Subject: Kaffe 0.9.2 broken for NEXTSTEP
In-Reply-To: <3437A28F.41C67EA6@tjwassoc.co.uk>
References: <3437A28F.41C67EA6@tjwassoc.co.uk>
Message-ID: <9710071248.AA27266@mathi.uni-heidelberg.DE>

Kaffe 0.9.2 is broken for all NEXTSTEP architectures in various aspects, =
it won't compile (shared libs are inherently broken, trampoline code =
doesn't compile etc.pp.).


I thought I had sent Tim a NOTES/README.NEXTSTEP for inclusion in 0.9.2 =
that points out those problems and I sent in some patches, but somehow =
they didn't show up in 0.9.2.


Therefore I would opt to remove NEXTSTEP from the list of supported =
platforms for the time. Perhaps some future release can fix and =
re-install it.


For what it's worth, this is my page dedicated to kaffe on =
NEXTSTEP/OPENSTEP. You find working source and binary distributions of =
kaffe 0.8.4 and kaffe 0.9.0 for NEXTSTEP Intel, SPARC and m68k:


  http://www.mathi.uni-heidelberg.de/~flight/STEPkaffe


	Gregor



---

|
<smaller> Gregor Hoffleit                        Mathematisches =
Institut, Uni HD |

| flight@mathi.uni-heidelberg.de      INF 288, 69120 Heidelberg, Germany =
|

| (NeXTmail, MIME)                      (49)6221 54-5771    fax  54-8312 =
|

| PGP Key fingerprint =3D 23 8F B3 38 A3 39 A6 01  5B 99 91 D6 F2 AC CD =
C7 |

</smaller>

From kaffe@kaffe.org  Tue Oct  7 17:31:20 1997
From: kaffe@kaffe.org (Artur Biesiadowski)
Date: Tue, 7 Oct 1997 18:31:20 +0200 (MET DST)
Subject: Construction of primitive classes
Message-ID: <Pine.GSO.3.96L01AT.971007182047.20433A-100000@sunrise>

I'm trying to implement some of the missing native funs for
java.lang.reflect.
My question is about creating objects of one of the primitive classes -
Integer, Double, Boolean etc. 
I could created them by calling
execute_java_constructor( many args...)
but this would make quite a lot overhead - all this varargs and checks for
illegal access.
Second options is just to call  
newObject( classptr );
and the set value manually - but this could crash if class is not loaded??
Question: are primitive classes guaranteed to be loaded/compiled at this
time - calls for Field.get/Field.set are not supposed to be taken at
initialization I think.
If not, will it be enough just to check class->state and process it in
case it is not already ? 
Or just forget about speedup and stick to safe vararg constructors ?


Artur

From kaffe@kaffe.org  Tue Oct  7 17:44:08 1997
From: kaffe@kaffe.org (Vik Sohal)
Date: Tue, 07 Oct 1997 09:44:08 -0700
Subject: Object Seralistion!
Message-ID: <2.2.32.19971007164408.00c1df40@wizard.lynx.com>

Object Serialization requires an understanding of the low level protocol
that SUN uses in their own implementation. Up till a week ago this protocol
was not published, but I understand that it is now. We are looking at it and
are thinking about implementing it in Kaffe. If we do, I am quite certain we
will be able to post it as a patch to the Kaffe community.

Vik Sohal

At 11:34 PM 10/6/97 +0100, you wrote:
>P. V. Thatayya Naidu wrote:
>> 
>> Hi,
>>     Will  Object Serialistion and RMI work with this new release Kaffe
0.9.2 ?
>
>I'm afraid not.  Any takers?
>
>Cheers
>Tim
>--
>  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
>  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
>  London, UK.                           Email:   tim@tjwassoc.co.uk
>
>
>
***************************************************************
* Vik Sohal                     *  email: vik@lynx.com        *
* Lynx Real-Time Systems        *  voice: (408) 879-3900 x119 *
* 2239 Samaritan Drive          *    fax: (408) 879-3920      *
* San Jose, CA 95124            *                             *
* ------------------------------*-----------------------------*
* cute msg: "We just turned it on, and it's I.Q. is already   *
*             in the thousands!"                              *
***************************************************************

From kaffe@kaffe.org  Tue Oct  7 18:26:57 1997
From: kaffe@kaffe.org (Per Bothner)
Date: Tue, 07 Oct 1997 10:26:57 -0700
Subject: Object Seralistion!
In-Reply-To: Your message of "Tue, 07 Oct 1997 09:44:08 PDT."
 <2.2.32.19971007164408.00c1df40@wizard.lynx.com>
Message-ID: <199710071726.KAA00461@cygnus.com>

> Up till a week ago this protocol was not published,

It has been published for a while (with the Serialization "white paper"),
(though I did not realize it until someone pointed it out).

> but I understand that it is now.

Perhaps there is an improved description.

Thre are rumors that Sun is changing RMI to be more compatible with CORBA.
I don't know what this entails.  Even if they do, the old protocol will
still be useful for quite a while.

	--Per Bothner
Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner

From kaffe@kaffe.org  Tue Oct  7 23:27:30 1997
From: kaffe@kaffe.org (Alexandre Oliva)
Date: 07 Oct 1997 20:27:30 -0200
Subject: Installing kaffe on IRIX 5.2 with GNU as
Message-ID: <orafgl43al.fsf@sunsite.dcc.unicamp.br>

Hi there!

I had trouble for compiling kaffe 0.9.2 on IRIX 5.2 because GNU as was
used instead of /usr/bin/as.  The object-file threads.o produced by
GNU as wouldn't link, whereas the one produced by /usr/bin/as would.
I'd suggest that, for the particular case of IRIX, the Makefile
fragment contained a definition such as:

AS=/usr/bin/as

-- 
Alexandre Oliva
mailto:oliva@dcc.unicamp.br mailto:aoliva@acm.org
http://www.dcc.unicamp.br/~oliva
Universidade Estadual de Campinas, SP, Brasil
From kaffe@kaffe.org  Tue Oct  7 22:10:12 1997
From: kaffe@kaffe.org (Tim Wilkinson)
Date: Tue, 07 Oct 1997 22:10:12 +0100
Subject: Object Seralistion!
References: <2.2.32.19971007164408.00c1df40@wizard.lynx.com>
Message-ID: <343AA534.41C67EA6@tjwassoc.co.uk>

Vik Sohal wrote:
> 
> Object Serialization requires an understanding of the low level protocol
> that SUN uses in their own implementation. Up till a week ago this protocol
> was not published, but I understand that it is now. We are looking at it and
> are thinking about implementing it in Kaffe. If we do, I am quite certain we
> will be able to post it as a patch to the Kaffe community.

I've actually looked into this and found that the documentation for the
original JDK 1.1-beta contains more than enough information on how to
write Serialization.  You should bear in mind that most of the hard work
will be done in the Java code and only the native stubs need to be
written - and these don't worry about the format of Serialized data
anyway.  Of course, the actual function of these native methods isn't
documented so you have to guess ...  But then, thats how the rest of the
native stuff was written.

Cheers
Tim
--
  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
  London, UK.                           Email:   tim@tjwassoc.co.uk

From kaffe@kaffe.org  Tue Oct  7 11:04:03 1997
From: kaffe@kaffe.org (Ralf W. Stephan)
Date: Tue, 7 Oct 1997 11:04:03 +0100
Subject: Object Seralistion!
In-Reply-To: <3439677C.3F54BC7E@tjwassoc.co.uk>; from Tim Wilkinson on Mon, Oct 06, 1997 at 11:34:36PM +0100
References: <199710061721.KAA01323@faith> <3439677C.3F54BC7E@tjwassoc.co.uk>
Message-ID: <19971007110403.51428@ark.franken.de>

Tim wrote 
> P. V. Thatayya Naidu wrote:
> >     Will  Object Serialistion and RMI work with this new release Kaffe 0.9.2 ?
> I'm afraid not.  Any takers?

No, I won't bite yet, I'm still trying to understand the source.
Maybe someone can help me:

- how do you know which native functions are needed by Java's classes?
  do you just wait for LinkErrors or is there something like an API?
- many files in packages/tjwassoc... are machine generated --- from
  what input?
- can the JITC in principle take a bytecode call to a member func
  and instead direct it to a native implementaton of the class?
  Example: java.util.Vector: could this be speeded up by *directly*
  calling my own C implementation (possibly without bounds checking, e.g.)?
- is there a point in rewriting some classes to get better performance /
  free versions of them, or is this outside the kaffe project?

Thanks,
ralf
-- 
URL: http://rws.home.pages.de/

From kaffe@kaffe.org  Wed Oct  8 20:51:45 1997
From: kaffe@kaffe.org (Darrell Gallion)
Date: Wed, 08 Oct 1997 15:51:45 -0400
Subject: Configure for QNX ??
Message-ID: <343BE451.B0D6847F@cldx.com>

After running configure, make fails.
Memory model and size of int, long, .. are wrong. And as far as I know
the Watcom compiler has no 64 bit
longs.

Thanx in advance for any insight.

Darrell Gallion
dgallion@cldx.com


From kaffe@kaffe.org  Thu Oct  9 00:02:56 1997
From: kaffe@kaffe.org (Stewart Allen)
Date: Wed, 8 Oct 1997 19:02:56 -0400 (EDT)
Subject: cannot build 0.92 on linux 2.0.30
Message-ID: <Pine.LNX.3.95.971008185534.24163A-100000@flex.neuron.com>


 About 3 seconds into the build I get:


make[4]: Entering directory
`/local/hd1/home/stewart/build/kaffe-0.9.2/packages/tjwassoc.co.uk/APIcore/lib'
gcc -g -O2 -I. -I../../../../include -I./../../../../include
-I../../../../config -I./../../../../config -c -fPIC ./java.lang/Math.c
In file included from ./java.lang/Math.c:16:
../../../../config/config-math.h:24: #error "Need some form of remainder"
make[4]: *** [Math.o] Error 1
make[4]: Leaving directory
`/local/hd1/home/stewart/build/kaffe-0.9.2/packages/tjwassoc.co.uk/APIcore/lib'


 Which I traced down to the fact that Configure incorrectly determines
 that the system doesn't support "fmod" or "drem" (both of which are
 present). I can correct these settings in config.h at which point the
 entire thing builds until the kaffevm link which gives:


gcc -g -O2  -I. -I./../kaffevm -I../../config -I./../../config
-I../../include -I./../../include -o kaffe main.o -L../kaffevm -lkaffevm
../kaffevm/libkaffevm.so: undefined reference to `remainder'
make[2]: *** [kaffe] Error 1


 Any suggestions?

 -stewart-

From kaffe@kaffe.org  Thu Oct  9 02:56:24 1997
From: kaffe@kaffe.org (Glynn Clements)
Date: Thu, 9 Oct 1997 02:56:24 +0100
Subject: cannot build 0.92 on linux 2.0.30
In-Reply-To: <Pine.LNX.3.95.971008185534.24163A-100000@flex.neuron.com>
References: <Pine.LNX.3.95.971008185534.24163A-100000@flex.neuron.com>
Message-ID: <199710090156.CAA08267@cerise.sensei.co.uk>


Stewart Allen wrote:

>  About 3 seconds into the build I get:
> 
> 
> make[4]: Entering directory
> `/local/hd1/home/stewart/build/kaffe-0.9.2/packages/tjwassoc.co.uk/APIcore/lib'
> gcc -g -O2 -I. -I../../../../include -I./../../../../include
> -I../../../../config -I./../../../../config -c -fPIC ./java.lang/Math.c
> In file included from ./java.lang/Math.c:16:
> .../../../../config/config-math.h:24: #error "Need some form of remainder"
> make[4]: *** [Math.o] Error 1
> make[4]: Leaving directory
> `/local/hd1/home/stewart/build/kaffe-0.9.2/packages/tjwassoc.co.uk/APIcore/lib'
> 
> 
>  Which I traced down to the fact that Configure incorrectly determines
>  that the system doesn't support "fmod" or "drem" (both of which are
>  present). I can correct these settings in config.h at which point the
>  entire thing builds until the kaffevm link which gives:
> 
> 
> gcc -g -O2  -I. -I./../kaffevm -I../../config -I./../../config
> -I../../include -I./../../include -o kaffe main.o -L../kaffevm -lkaffevm
> .../kaffevm/libkaffevm.so: undefined reference to `remainder'
> make[2]: *** [kaffe] Error 1

This indicates that remainder isn't being replaced with fmod.

Make sure that HAVE_REMAINDER is undefined and HAVE_FMOD is defined,
e.g. insert

	#undef HAVE_REMAINDER
	#define HAVE_FMOD 1

at the top of config-math.h.

I'm running Linux 2.0.30 with libc 5.4.23, and it compiled out of the
box for me.

-- 
Glynn Clements <glynn@sensei.co.uk>

From kaffe@kaffe.org  Thu Oct  9 02:56:24 1997
From: kaffe@kaffe.org (Glynn Clements)
Date: Thu, 9 Oct 1997 02:56:24 +0100
Subject: cannot build 0.92 on linux 2.0.30
In-Reply-To: <Pine.LNX.3.95.971008185534.24163A-100000@flex.neuron.com>
References: <Pine.LNX.3.95.971008185534.24163A-100000@flex.neuron.com>
Message-ID: <199710090156.CAA08267@cerise.sensei.co.uk>


Stewart Allen wrote:

>  About 3 seconds into the build I get:
> 
> 
> make[4]: Entering directory
> `/local/hd1/home/stewart/build/kaffe-0.9.2/packages/tjwassoc.co.uk/APIcore/lib'
> gcc -g -O2 -I. -I../../../../include -I./../../../../include
> -I../../../../config -I./../../../../config -c -fPIC ./java.lang/Math.c
> In file included from ./java.lang/Math.c:16:
> .../../../../config/config-math.h:24: #error "Need some form of remainder"
> make[4]: *** [Math.o] Error 1
> make[4]: Leaving directory
> `/local/hd1/home/stewart/build/kaffe-0.9.2/packages/tjwassoc.co.uk/APIcore/lib'
> 
> 
>  Which I traced down to the fact that Configure incorrectly determines
>  that the system doesn't support "fmod" or "drem" (both of which are
>  present). I can correct these settings in config.h at which point the
>  entire thing builds until the kaffevm link which gives:
> 
> 
> gcc -g -O2  -I. -I./../kaffevm -I../../config -I./../../config
> -I../../include -I./../../include -o kaffe main.o -L../kaffevm -lkaffevm
> .../kaffevm/libkaffevm.so: undefined reference to `remainder'
> make[2]: *** [kaffe] Error 1

This indicates that remainder isn't being replaced with fmod.

Make sure that HAVE_REMAINDER is undefined and HAVE_FMOD is defined,
e.g. insert

	#undef HAVE_REMAINDER
	#define HAVE_FMOD 1

at the top of config-math.h.

I'm running Linux 2.0.30 with libc 5.4.23, and it compiled out of the
box for me.

-- 
Glynn Clements <glynn@sensei.co.uk>
From kaffe@kaffe.org  Thu Oct  9 10:10:55 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Thu, 9 Oct 1997 10:10:55 +0100
Subject: Configure for QNX ??
In-Reply-To: <343BE451.B0D6847F@cldx.com> (message from Darrell Gallion on
 Wed, 8 Oct 1997 20:51:45 +0100)
Message-ID: <9710091010.aa11291@stlind.lint.lyngso-industri.dk>


> After running configure, make fails.
> Memory model and size of int, long, .. are wrong. And as far as I know
> the Watcom compiler has no 64 bit
> longs.

Watcom 11 has 64 bit ints BUT you must use gcc for compiling kaffe on
QNX. 

Note that kaffe-0.9.2 does not compile out of the box for QNX. 

0.9.1, however, does with a little help. 

Here is the message that I posted back in july to comp.os.qnx,
describing how to compile kaffe-0.9.1 on QNX.

---------------------------------------------------------------------

I have ported the kaffe virtual machine to QNX-4.23. 

Kaffe is a free implementation of the Java virtual machine, written by
Tim Wilkinson. Kaffe uses Just-In-Time compilation of byte-code into
machine code which makes it pretty efficient (considerably faster than
Sun's "reference" interpreter based implementation - slightly slower
than Sun's JIT based implementation).

Kaffe can be used together with Sun's class libraries version
1.1.2. The complete API is not yet supported by native methods though
- I think more or less the subset corresponding to version 1.0.2 is
supported. An important exception is that awt (i.e., graphics) is not
supported by kaffe-0.9.1. 

Gnu CC 2.7.2 is required in order to compile kaffe. gcc can be
obtained from http://www.teaser.fr/~jcmichot/gcc.

The QNX patches have already been incorporated into kaffe-0.9.1, which
can be obtained from ftp://ftp.kaffe.org/pub/kaffe/kaffe-0.9.1.tgz. 

A single bugfix is required to make it compile: in
config/i386/trampolines.c replace ".global" with ".globl"

Sun's class libraries are available from a number of places, either as
part of kaffe (kaffe-0.9.1-package-javasoft.tgz), or as part of
jdk-1.1.2 which can be downloaded from javasoft.com.


Regards, Jeppe Sommer
jso@lyngso-industri.dk

----------------------------------------------------------------------

One more note: the ld command provided with gcc will confuse kaffe's
configure script because it exits zero even when unresolved symbols
are encountered. A fixed (beta) ld can be obtained from 
http://www.teaser.fr/~jcmichot/gcc/ld. I suspect that this will be a
temporary issue. 



Jeppe

From kaffe@kaffe.org  Thu Oct  9 21:42:19 1997
From: kaffe@kaffe.org (Alain Magloire)
Date: Thu, 9 Oct 1997 16:42:19 -0400 (EDT)
Subject: {K/J}NI
In-Reply-To: <3438E1BB.1CFBAE39@tjwassoc.co.uk> from "Tim Wilkinson" at Oct 6, 97 02:03:55 pm
Message-ID: <199710092042.QAA01026@pct62.cae.ca>

Bonjour M. Wilkinson

  Is kaffe supporting JNI, if not is there plan for it ?

I see that kaffe uses the old stub interface 

Thanks


--
alain

From kaffe@kaffe.org  Fri Oct 10 20:59:05 1997
From: kaffe@kaffe.org (Amitabh Arora)
Date: Fri, 10 Oct 1997 12:59:05 -0700
Subject: kaffe port for HP-UX 9.05 OS
Message-ID: <01BCD57C.4AF62E80@amitabh@aspec.com>

Has anyone done this port?  I do not need the JIT feature.
So the port should be easy.

Thanks. -amitabh arora
From kaffe@kaffe.org  Sat Oct 11 00:34:46 1997
From: kaffe@kaffe.org (James A. Hillyerd)
Date: Fri, 10 Oct 1997 16:34:46 -0700 (PDT)
Subject: JavaLobby and JavaOS
Message-ID: <Pine.LNX.3.95.971010163139.30292A-100000@animal.blarg.net>

There is some interest on the JavaLobby to write a JavaOS.  This would
probably be based on the Mach microkernel, and perhaps some pieces of GNU
Hurd.  The thought was that Kaffe would be embedded into this.  The
working name seems to be 'Oz': as in the land where anything can happen.

We could really use some C hackers, and some Kernel hackers.  Please join
in the discussion, and you should also consider joining the Java Lobby,
it's free!

http://www.javalobby.com/

Sorry if this irritates anyone.  Nobody asked me to post this here, so
direct all flames to me. =)

-james

[  James A. Hillyerd (JH2162) - james@blarg.net - Web Developer  ]
[  http://www.blarg.net/~james/   http://www.hyperglyphics.com/  ]
[ 1024/B11C3751 CA 1C B3 A9 07 2F 57 C9  91 F4 73 F2 19 A4 C5 88 ]

From kaffe@kaffe.org  Sat Oct 11 12:42:32 1997
From: kaffe@kaffe.org (Fletcher E Kittredge)
Date: Sat, 11 Oct 1997 07:42:32 -0400
Subject: JavaLobby and JavaOS
In-Reply-To: Your message of "Fri, 10 Oct 1997 16:34:46 PDT."
 <Pine.LNX.3.95.971010163139.30292A-100000@animal.blarg.net>
Message-ID: <199710111142.HAA14774@river.gwi.net>

On Fri, 10 Oct 1997 16:34:46 -0700 (PDT)  "James A. Hillyerd" wrote:
> There is some interest on the JavaLobby to write a JavaOS.  This would
> probably be based on the Mach microkernel, and perhaps some pieces of GNU
> Hurd.  The thought was that Kaffe would be embedded into this.  The
> working name seems to be 'Oz': as in the land where anything can happen.
> 
> We could really use some C hackers, and some Kernel hackers.  Please join
> in the discussion, and you should also consider joining the Java Lobby,
> it's free!

As an ex-Mach and C hacker, let me say that there are real good
technical reasons that Mach and Hurd failed.  The micro-kernel
architecture is dog slow; adding Java to the mix is not going to speed
it up.

Before wasting good Java developer time on this, please have someone
you trust go back and read all the papers from the late 80's, early
90's which detail the significant performance problems with a
micro-kernel architecture...

regards,
fletcher

From kaffe@kaffe.org  Sat Oct 11 14:26:05 1997
From: kaffe@kaffe.org (Vijay Saraswat)
Date: Sat, 11 Oct 1997 09:26:05 -0400 (EDT)
Subject: JavaLobby and JavaOS
In-Reply-To: "James A. Hillyerd"'s message of Fri, 10 Oct 1997 16:34:46 -0700 (PDT)
Message-ID: <199710111326.JAA23400@radish.research.att.com>

Oz is the name of a programming system produced by the German AI
institute (DFKI). 

From kaffe@kaffe.org  Sat Oct 11 15:51:01 1997
From: kaffe@kaffe.org (Sean McDirmid)
Date: Sat, 11 Oct 1997 07:51:01 -0700 (PDT)
Subject: JavaLobby and JavaOS
In-Reply-To: <199710111142.HAA14774@river.gwi.net>
Message-ID: <Pine.OSF.3.96.971011074458.4584A-100000@sanjuan.cs.washington.edu>

On Sat, 11 Oct 1997, Fletcher E  Kittredge wrote:

> As an ex-Mach and C hacker, let me say that there are real good
> technical reasons that Mach and Hurd failed.  The micro-kernel
> architecture is dog slow; adding Java to the mix is not going to speed
> it up.
> 
> Before wasting good Java developer time on this, please have someone
> you trust go back and read all the papers from the late 80's, early
> 90's which detail the significant performance problems with a
> micro-kernel architecture...

Now hold on here, if I understand it, the micro-kernal architecture was
slow b/c of the way it kept its form of modularity, through runtime
barriers and abstractions.  The great thing about Java code is that it is
safe and this safety can be shown (lacking in kaffe but I've written a
verifier for kimera - kimera.cs.washington.edu).  Several good performing
(and commercial) operating systems have been "derived" from
micro-kernal's, such as NT, NextStep and perhaps the upcoming BeOS. How
about a derived micro-kernal that takes Java code, compiles it, and places
it close to the micro-kernal as a trusted extension?  Now we don't have a
micro-kernal anymore but you might be able to get the performance you are
looking for. You still get the safety and abstraction that the original
micro-kernal provided (through the virtual machine compiled to machine
code interface).

Sean McDirmid
kimera.cs.washington.edu


From kaffe@kaffe.org  Sat Oct 11 17:28:18 1997
From: kaffe@kaffe.org (Bruce R. Montague)
Date: Sat, 11 Oct 1997 09:28:18 -0700
Subject: JavaLobby and JavaOS
Message-ID: <199710111628.JAA02908@madrone.cse.ucsc.edu>

 re microkernel OS engineering/performance problems - 2 recent papers 
 that may prove of interest in this regard are:

 "Experience with the Development of a Microkernel-Based, Multiserver
  Operating System", Freeman Rawson, Proc. 6th Workshop on Hot Topics
  in Operating Systems, 5-6 May 1997 (HOTOS);
 
 "The Failure of Personalities to Generalize", Brett Fleisch, ditto...

 These are also intrinsically of interest as IBM apparently lost
 $2 _billion_ on this... Quote: "Despite the level of investment and the
 adoption of what was deemed to be the best of operating systems research
 and advanced development of the 1980s, the project, in retrospect, was
 not a technical success." (Rawson).

 Apologies if this is off subject, I'm just seconding the motion to 
 think this through carefully; forewarned is forarmed, ... investing
 the effort in things like porting interfaces, validation/compatibility
 tests, etc., might be more effective?

 Regards,
 bruce

From kaffe@kaffe.org  Sat Oct 11 17:55:45 1997
From: kaffe@kaffe.org (Markus Peter)
Date: Sat, 11 Oct 1997 18:55:45 +0200
Subject: JavaLobby and JavaOS
Message-ID: <01bcd666$849cd450$0201a8c0@warpnt.spin.de>

Ok. I guess it's also time for a statement from you.
I was one of the initiators of this project.

In the mean time, we already have found somebody,
who will create a new kernel specifically for a Java OS.
This one will be very small, as a Java based OS eliminates
a lot of needs (process switching, protected memory etc...)

As soon as we have that OS, we will need a port of Kaffe
for it - I've chosen Kaffe as JIT Compiler for this OS,
because it's free and because I once needed a JIT Compiler
for Linux, and I was very disappointed of the Blackdown port,
while Kaffe appeared very excellent to me (even at that time
- it was version 0.8.4).

Unfortunately the person who will create our OS cannot
port Kaffe himself, as he's got a Java source code license
from Sun, and we want a Java OS free from any copyrights
ASAP.

That's where we will need some help from one of you for sure.
Most of the volunteers right now are experienced Java programmers,
but they do not have enough C experience to port Kaffe or it
is too long ago they used C. So if one of you is interested in
this project and would be willing to port kaffe to that OS kernel,
as soon as it is created, please contact me.

I guess without the help of one of you this project is determined
to die.

(Hehe - please do not flame me with telling me, that I should
not try to create an OS without having a clue of it, in the first place :)

Markus Peter (Developer), SPiN GmbH
warpi@spin.de   -    http://www.spin.de
Java, Perl and Web Site Development
Unite for Java!  - http://www.javalobby.org/


From kaffe@kaffe.org  Sat Oct 11 18:25:05 1997
From: kaffe@kaffe.org (roy)
Date: Sat, 11 Oct 1997 10:25:05 -0700
Subject: JavaLobby and JavaOS
In-Reply-To: <Pine.OSF.3.96.971011074458.4584A-100000@sanjuan.cs.washing
 ton.edu>
References: <199710111142.HAA14774@river.gwi.net>
Message-ID: <3.0.3.32.19971011102505.009652e0@mail.rdc1.sfba.home.net>


A microkernel is probably the worst possible fit for a 
"pure Java OS" (something that supports only Java)
if you are going to maintain the microkernel view of
things and run Kaffe as a server or subsystem.

Due to Java's security and safety, a conventional OS
carries a lot of extra baggage that is unnecessary for
a "pure Java OS".
 
The objective of microkernels is to isolate and
optimize mechanism in a small core, while providing
for multiple policies in user space. But with a 
"pure Java OS" this is unnecessary, since all you
need to support is Java. Furthermore, with a verifier
to assist, you may not need a user/kernel distinction
any more.

Instead what one probably should do is link Kaffe directly
into the kernel of some free OS (preferably real-time),
then throw out all the unnecessary stuff (user-kernel
distinction, processes, etc...)

A number of commercial and research OS vendors have used
this approach (including Chorus and the Univ. of Utah. Except
that neither is real-time, which may be important for us.)
i.e. to turning a conventional OS into a "pure java OS"
involves mostly paring stuff away.

Come to think of it, a micro-kernel may be a good starting
point if we were to link Kaffe directly into it (instead of
making it a "single server" or a "subsystem"), then turn
the non-Kaffe portions into a "pico-kernel" :-)

It may be that single-address space OSs are the best
match for Java. Perhaps we should look at Opal. Or a
kernel-extensible OS like Spin. These would probably
a much better starting point than Mach or Hurd.

By the way, I think Next is based on Mach 2.5, which is
not a microkernel. Neither is the current release of NT.

Roy


From kaffe@kaffe.org  Sat Oct 11 19:18:15 1997
From: kaffe@kaffe.org (Markus Peter)
Date: Sat, 11 Oct 1997 20:18:15 +0200
Subject: JavaOS
Message-ID: <01bcd672$0abc9c40$0201a8c0@warpnt.spin.de>

I tried emailing this mailing list earlier, but strangely, my message
did not arrive - well - here it is again - I hope you now do not
get two copies of it.

Ok. I guess it's also time for a statement from me.
I was one of the initiators of this project.

In the mean time, we already have found somebody,
who will create a new kernel specifically for a Java OS.
This one will be very small, as a Java based OS eliminates
a lot of needs (process switching, protected memory etc...)

As soon as we have that OS, we will need a port of Kaffe
for it - I've chosen Kaffe as JIT Compiler for this OS,
because it's free and because I once needed a JIT Compiler
for Linux, and I was very disappointed of the Blackdown port,
while Kaffe appeared very excellent to me (even at that time
- it was version 0.8.4).

Unfortunately the person who will create our OS cannot
port Kaffe himself, as he's got a Java source code license
from Sun, and we want a Java OS free from any copyrights
ASAP.

That's where we will need some help from one of you for sure.
Most of the volunteers right now are experienced Java programmers,
but they do not have enough C experience to port Kaffe or it
is too long ago they used C. So if one of you is interested in
this project and would be willing to port kaffe to that OS kernel,
as soon as it is created, please contact me.
I guess without the help of one of you this project is determined
to die.

(Hehe - please do not flame me with telling me, that I should
not try to create an OS without having a clue of it, in the first place :)

And one last note to Bruce Montague:
Hurd did not fail yet - in fact the first versions of Hurd were released
some months ago.
And yes - I know, lots of people say, microkernel designs have failed,
but on the other hand there are also papers citing that they are useful.
WinNT uses the MachKernel, NextStep does, Rhapsody will do.
But that does not matter anymore - after all we now have found
a developer who will create an exclusive kernel for us.

I wanted to use a microkernel in the beginning, because most
people participating in the project at the beginning were only Java
developers but were not very experienced with C programming.
So I though we should take something existing, maybe a micro-kernel,
because it is not bloated too much with unnecessary stuff.
After all a true kernel for a Java OS need not do much.


Markus Peter (Developer), SPiN GmbH
warpi@spin.de   -    http://www.spin.de
Java, Perl and Web Site Development
Unite for Java!  - http://www.javalobby.org/

From kaffe@kaffe.org  Sun Oct 12 09:37:25 1997
From: kaffe@kaffe.org (Gabriel Sebestyen)
Date: Sun, 12 Oct 1997 10:37:25 +0200
Subject: JavaOS
References: <01bcd672$0abc9c40$0201a8c0@warpnt.spin.de>
Message-ID: <34408C45.3041166C@accentcomm.com>

Hi,

First of all, Windows NT(tm :-) ) uses an own kernel developed by DEC
not Mach. Mach is designed to be a small and independent microkernel
which can be used to build your own system on. Mach 3 finally lost the
only (BSD) UN*X dependent part now it has a BSD server instead.
Some facts about Mach:
* unified and VERY efficient message handling
* simple but efficient virtual memory handling (who likes NT's memory
management? :-) )
* supports threads
* ported to a lot of platforms but it's as independent from them as it
can be

I think Mach is a real microkernel (because it was designed to be
originally and choosen by a lot of OS's) and an excellent starting point
to develope e.g. a JavaOS. Making a new kernel from the beginning does
not make sense! Just a little performance example:
An HP UNIX emulator to Mach wrote by a team for test purposes was faster
than the original HP! It's true believe me!

Gabriel Sebestyen


From kaffe@kaffe.org  Mon Oct 13 07:57:34 1997
From: kaffe@kaffe.org (Luis L. Fernandez)
Date: Mon, 13 Oct 1997 08:57:34 +0200
Subject: JavaOS
Message-ID: <199710130657.IAA00386@diogenes.iit.upco.es>


Hi,

	
	I've listening your discussion about a new Java-OS and
micro-kernels design....

	Well, in the Computer Science Department at Utah University
you can find the answer to this discussion. They have already
developed a Java-OS using a library with provides you a toolkit to
build OS named OSkit upon and Intel ix86 architecture. I've used it
and it's really good. The OSKit code it's free and it's available in

http://www.cs.utah.edu/projects/flux/oskit/

They are kernel hackers and they have redesigned the Mach micro-kernel
to run fine (search about Fluke project).

I hope this message seems useful to somebody ;-)

Best Regards,

	Luis.

---------
Luis Lopez Fernandez
e-mail: llf@computer.org

From kaffe@kaffe.org  Mon Oct 13 12:17:31 1997
From: kaffe@kaffe.org (Vijay Saraswat)
Date: Mon, 13 Oct 1997 07:17:31 -0400 (EDT)
Subject: JavaLobby and JavaOS
In-Reply-To: Sean McDirmid's message of Sat, 11 Oct 1997 07:51:01 -0700 (PDT)
Message-ID: <199710131117.HAA03790@radish.research.att.com>

   X-Authentication-Warning: vega.soi.city.ac.uk: lp set sender to owner-kaffe@soi.city.ac.uk using -f
   From: Sean McDirmid <mcdirmid@cs.washington.edu>
   Sender: owner-kaffe@soi.city.ac.uk

   On Sat, 11 Oct 1997, Fletcher E  Kittredge wrote:

   > As an ex-Mach and C hacker, let me say that there are real good
   > technical reasons that Mach and Hurd failed.  The micro-kernel
   > architecture is dog slow; adding Java to the mix is not going to speed
   > it up.
   > 
   > Before wasting good Java developer time on this, please have someone
   > you trust go back and read all the papers from the late 80's, early
   > 90's which detail the significant performance problems with a
   > micro-kernel architecture...

   Now hold on here, if I understand it, the micro-kernal architecture was
   slow b/c of the way it kept its form of modularity, through runtime
   barriers and abstractions.  The great thing about Java code is that it is
   safe and this safety can be shown (lacking in kaffe but I've written a
   verifier for kimera - kimera.cs.washington.edu).  Several good performing
   (and commercial) operating systems have been "derived" from
   micro-kernal's, such as NT, NextStep and perhaps the upcoming BeOS. How
   about a derived micro-kernal that takes Java code, compiles it, and places
   it close to the micro-kernal as a trusted extension?  Now we don't have a
   micro-kernal anymore but you might be able to get the performance you are
   looking for. You still get the safety and abstraction that the original
   micro-kernal provided (through the virtual machine compiled to machine
   code interface).

I agree with Sean here. The notion of a strongly-typed,
dynamically linked language, like Java, makes some fresh
approaches possible. 

To go a step beyond what Sean is saying: in fact it is possible
(the "declarative bytecode verification" approach) to develop
bytecode verifiers *systematically*, using a first-principles
approach, from a constraint-based specification of the opcodes
involved. One can translate an input byte code program, P,
compositionally (byte-code for byte-code) into a constraint
program, T(P) which "obviously" states the constraints on the
run-time type-state of the Java Virtual Machine, which must hold
for program execution to be type-safe. T(P) is "loop-free" and
will take time proportional to its length, which is proportional
to the length of P, to execute. If T(P) does not deadlock or
produce "false", P is type-safe.

This approach differs slightly from the one that Sean and
Javasoft have implemented. In their cases, there are separate
*byte code verifiers*, usually largish (10-15 page) C programs,
implementing a "data-flow" analysis, documented by 10-15 pages of
informal English (I have not seen Sean's documentation yet),
about whose correctness it is hard to say anything formal. 

In any case, the bottom line for this list should be, I think,
that one can assume that the subproblem of verifying the
type-safety of JVM-like bytecode is solved, with of course, one
important caveat: native code. (A lot of the "core" classes in
Java in fact contain natively implemented methods, and one has to
use separate techniques for verifying them.) 

I would not be too surprised if this approach (Java's approach)
of doing "link-time" (i.e. first-instruction-execution-time)
type-checking actually resulted in significant performance
*improvements* over standard OS's. One thing to keep in mind is
that Java's architecture is such that the bytecode verifier can
in fact accept some input code *even if it is not type-safe in
isolation* --- as long as it is type-safe when combined with the
other classes that are already loaded. That is, there are some
possibilities of doing inter-classfile optimizations (e.g.
eliminating some classcasts generated by the compiler) in a
typesafe way, that can produce even more performance benefits
than are posible if one does "straight" Java verification, as
done currently by Sun's Java systems.

Best,
Vijay

From kaffe@kaffe.org  Mon Oct 13 12:10:10 1997
From: kaffe@kaffe.org (Roger Binns)
Date: Mon, 13 Oct 1997 11:10:10 +0000 (GMT)
Subject: JavaLobby and JavaOS
In-Reply-To: <01bcd666$849cd450$0201a8c0@warpnt.spin.de> from "Markus Peter" at Oct 11, 97 06:55:45 pm
Message-ID: <199710131110.LAA25110@avon.x.co.uk>

> In the mean time, we already have found somebody,
> who will create a new kernel specifically for a Java OS.
> This one will be very small, as a Java based OS eliminates
> a lot of needs (process switching, protected memory etc...)

One thing to be careful of is drivers.  I don't know if you plan
to support commodity PC hardware, or will have very few supported
devices.

In any case, my suggestion would be to take a freeware UNIX (eg Linux)
that has lots of device support, and slowly remove/rewrite bits
to get to your goal.  This has the advantage that you have a working system 
from day one.  

(Ironically, it would also mirror the way Linux was built from Minix!)

Roger
-- 
Roger Binns   rogerb@sco.com | The reasonable man adapts himself to the world;
Software Architect           | the unreasonable one persists in trying to
Client Integration Division  | to adapt the world to himself. Therefore all
SCO, Vision Park, Cambridge  | progress depends on the unreasonable man - GBS

From kaffe@kaffe.org  Mon Oct 13 15:12:20 1997
From: kaffe@kaffe.org (Aleksey Sudakov)
Date: Mon, 13 Oct 97 10:12:20 -0400
Subject: JavaOS
In-Reply-To: <199710130657.IAA00386@diogenes.iit.upco.es>
References: <199710130657.IAA00386@diogenes.iit.upco.es>
Message-ID: <9710131412.AA07422@rkinc.com>

: 	Well, in the Computer Science Department at Utah University
: you can find the answer to this discussion. They have already
: developed a Java-OS using a library with provides you a toolkit to
: build OS named OSkit upon and Intel ix86 architecture. I've used it
: and it's really good. The OSKit code it's free and it's available in

It's totally amazing. The only bad things are that it's not available for  
immediate download and it's not free. They say that it's free for  
non-commercial use and IMHO it will affect acceptance of this totally cool  
idea.

I wonder when free software become really free.

Regards,
Aleksey

From kaffe@kaffe.org  Mon Oct 13 15:29:50 1997
From: kaffe@kaffe.org (Vijay Saraswat)
Date: Mon, 13 Oct 1997 10:29:50 -0400 (EDT)
Subject: JavaLobby and JavaOS
In-Reply-To: "Bruce R. Montague"'s message of Sat, 11 Oct 1997 09:28:18 -0700
Message-ID: <199710131429.KAA05876@radish.research.att.com>

RE: the HOTOS talks, a URL is

  http://www.eecs.harvard.edu/hotos/scribes/sessions/implementation-1.html


   Date: Sat, 11 Oct 1997 09:28:18 -0700
   From: "Bruce R. Montague" <brucem@cse.ucsc.edu>

    re microkernel OS engineering/performance problems - 2 recent papers 
    that may prove of interest in this regard are:

    "Experience with the Development of a Microkernel-Based, Multiserver
     Operating System", Freeman Rawson, Proc. 6th Workshop on Hot Topics
     in Operating Systems, 5-6 May 1997 (HOTOS);

    "The Failure of Personalities to Generalize", Brett Fleisch, ditto...

From kaffe@kaffe.org  Mon Oct 13 16:01:38 1997
From: kaffe@kaffe.org (Thanh Ma)
Date: Mon, 13 Oct 1997 11:01:38 -0400
Subject: JavaOS
In-Reply-To: <9710131412.AA07422@rkinc.com>; from Aleksey Sudakov on Mon, Oct 13, 1997 at 10:12:20AM -0400
References: <199710130657.IAA00386@diogenes.iit.upco.es> <9710131412.AA07422@rkinc.com>
Message-ID: <19971013110137.07019@dino.encore.com>

On Mon, Oct 13, 1997 at 10:12:20AM -0400, Aleksey Sudakov wrote:
> 
> It's totally amazing. The only bad things are that it's not available for  
> immediate download and it's not free. They say that it's free for  
> non-commercial use and IMHO it will affect acceptance of this totally cool  
> idea.
> 
> I wonder when free software become really free.

SCO's commercial UnixWare and OpenServer are 'free' for non-commercial
purposes :-)

Thanh

From kaffe@kaffe.org  Mon Oct 13 16:38:00 1997
From: kaffe@kaffe.org (Patrick Logan)
Date: Mon, 13 Oct 97 08:38 PDT
Subject: JavaLobby and JavaOS
In-Reply-To: <Pine.LNX.3.95.971010163139.30292A-100000@animal.blarg.net>
References: <Pine.LNX.3.95.971010163139.30292A-100000@animal.blarg.net>
Message-ID: <m0xKmZG-000DM5C@gharlane.gemstone.com>

>>>>> "James" == James A Hillyerd <james@blarg.net> writes:

    James> There is some interest on the JavaLobby to write a JavaOS.
    James> This would probably be based on the Mach microkernel, and
    James> perhaps some pieces of GNU Hurd.

Someone from Utah already ported kaffe to their OS framework. This
framework was developed by former Mach developers to *replace* Mach
because it was too unwieldy. Check it out in the research pages at
www.cs.utah.edu, I believe.

-- 
Patrick Logan                 mailto:patrickl@gemstone.com
Voice 503-533-3365            Fax   503-629-8556
Gemstone Systems, Inc         http://www.gemstone.com

From kaffe@kaffe.org  Mon Oct 13 17:02:00 1997
From: kaffe@kaffe.org (Mark Huizer)
Date: Mon, 13 Oct 1997 18:02:00 +0200
Subject: JavaOS
In-Reply-To: <9710131412.AA07422@rkinc.com>; from "Aleksey Sudakov" on Mon, Oct 13, 1997 at 10:12:20AM -0400
References: <199710130657.IAA00386@diogenes.iit.upco.es> <9710131412.AA07422@rkinc.com>
Message-ID: <19971013180200.04495@pcnov032.win.tue.nl>

The wise Aleksey Sudakov produced the following lines:
> : 	Well, in the Computer Science Department at Utah University
> : you can find the answer to this discussion. They have already
> : developed a Java-OS using a library with provides you a toolkit to
> : build OS named OSkit upon and Intel ix86 architecture. I've used it
> : and it's really good. The OSKit code it's free and it's available in
> 
> It's totally amazing. The only bad things are that it's not available for  
> immediate download and it's not free. They say that it's free for  
> non-commercial use and IMHO it will affect acceptance of this totally cool  
> idea.
> 
Can you give me some URL or mailaddress so I can look at it? Sounds
cool, and yes, I'm not commercial planning to use it :-)

Mark

From kaffe@kaffe.org  Mon Oct 13 17:43:00 1997
From: kaffe@kaffe.org (Patrick Logan)
Date: Mon, 13 Oct 97 09:43 PDT
Subject: JavaOS
In-Reply-To: <9710131412.AA07422@rkinc.com>
References: <199710130657.IAA00386@diogenes.iit.upco.es>
 <9710131412.AA07422@rkinc.com>
Message-ID: <m0xKnaj-000DM5C@gharlane.gemstone.com>

>>>>> "Aleksey" == Aleksey Sudakov <zander@rkinc.com> writes:

    Aleksey> ...They say that it's free for non-commercial use...
    Aleksey> I wonder when free software become really free.

Why do you expect to make a profit from other people's labors? Isn't
it fair that, if you are going to charge money for something that you
didn't do, that you give a portion of that money to the ones who did
the work?

(Insert story of Henny Penny here.)

Look at it this way, giving money back to the ones who did the work
will keep *them* in business, so they can do *more* work.

Enough said in this list, please mail me personally about
disagreements.

-- 
Patrick Logan                 mailto:patrickl@gemstone.com
Voice 503-533-3365            Fax   503-629-8556
Gemstone Systems, Inc         http://www.gemstone.com

From kaffe@kaffe.org  Mon Oct 13 18:17:27 1997
From: kaffe@kaffe.org (Bartlee A. Anderson)
Date: Mon, 13 Oct 1997 12:17:27 -0500
Subject: JavaOS
References: <199710130657.IAA00386@diogenes.iit.upco.es>
Message-ID: <344257A7.1B4@switch.rockwell.com>

Luis,
I saw a reference to OSkit 5 months ago and its still not available
at the site you posted. What's the deal?
Bart.
Luis L. Fernandez wrote:
> 
> Hi,
> 
> 
>         I've listening your discussion about a new Java-OS and
> micro-kernels design....
> 
>         Well, in the Computer Science Department at Utah University
> you can find the answer to this discussion. They have already
> developed a Java-OS using a library with provides you a toolkit to
> build OS named OSkit upon and Intel ix86 architecture. I've used it
> and it's really good. The OSKit code it's free and it's available in
> 
> http://www.cs.utah.edu/projects/flux/oskit/
> 
> They are kernel hackers and they have redesigned the Mach micro-kernel
> to run fine (search about Fluke project).
> 
> I hope this message seems useful to somebody ;-)
> 
> Best Regards,
> 
>         Luis.
> 
> ---------
> Luis Lopez Fernandez
> e-mail: llf@computer.org

-- 
Bartlee A. Anderson     |  Sys. Test (ISDN-Commands-Trans Link-Tools)
Rockwell International  |                Electronic Commerce Division
300 Bauman Ct.          |                    banders@ecd.rockwell.com
MS 933-605      Opinions my own, not Rockwell's  VOICE (630) 227-8975
Wood Dale, IL 60191     |                        FAX   (630) 227-9771

From kaffe@kaffe.org  Mon Oct 13 18:24:11 1997
From: kaffe@kaffe.org (Luis L. Fernandez)
Date: Mon, 13 Oct 1997 19:24:11 +0200
Subject: JavaOS
In-Reply-To: <19971013180200.04495@pcnov032.win.tue.nl>
References: <199710130657.IAA00386@diogenes.iit.upco.es>
 <9710131412.AA07422@rkinc.com>
 <19971013180200.04495@pcnov032.win.tue.nl>
Message-ID: <199710131724.TAA02484@diogenes.iit.upco.es>

Mark Huizer writes:
 > > It's totally amazing. The only bad things are that it's not available for  
 > > immediate download and it's not free. They say that it's free for  
 > > non-commercial use and IMHO it will affect acceptance of this totally cool  
 > > idea.
 > > 
 > Can you give me some URL or mailaddress so I can look at it? Sounds
 > cool, and yes, I'm not commercial planning to use it :-)
 >

Sure, you can find the OSkit at this site:

	http://www.cs.utah.edu/projects/flux/oskit/

Enjoy!!

o-------------------------------o------------------------------------o
| Luis Lopez Fernandez          |  e-mail: luisl@iit.upco.es         |   
| Industrial Systems Division   |          llf@computer.org          |
o-------------------------------o------------------------------------o
|              Technological Research Institute                      |
|          Santa Cruz de Marcenado, 26 - Planta 1 -                  |
|	            28015 Madrid - SPAIN                             |
|      Tel: 34-1-5242800 (ext. 2706) - FAX: 34-1-5423176             |
o--------------------------------------------------------------------o

From kaffe@kaffe.org  Mon Oct 13 18:42:49 1997
From: kaffe@kaffe.org (Dan Lambright)
Date: Mon, 13 Oct 1997 13:42:49 -0400
Subject: JavaLobby and JavaOS
In-Reply-To: Your message of "Mon, 13 Oct 1997 07:17:31 EDT."
 <199710131117.HAA03790@radish.research.att.com>
Message-ID: <199710131742.NAA00507@postman.opengroup.org>

In message <199710131117.HAA03790@radish.research.att.com>Vijay Saraswat writes
>   X-Authentication-Warning: vega.soi.city.ac.uk: lp set sender to owner-kaffe
>@soi.city.ac.uk using -f
.
.
.
>I agree with Sean here. The notion of a strongly-typed,
>dynamically linked language, like Java, makes some fresh
>approaches possible. 
>
>To go a step beyond what Sean is saying: in fact it is possible
>(the "declarative bytecode verification" approach) to develop
>bytecode verifiers *systematically*, using a first-principles
>approach, from a constraint-based specification of the opcodes
>involved. One can translate an input byte code program, P,
>compositionally (byte-code for byte-code) into a constraint
>program, T(P) which "obviously" states the constraints on the
>run-time type-state of the Java Virtual Machine, which must hold
>for program execution to be type-safe. T(P) is "loop-free" and
>will take time proportional to its length, which is proportional
>to the length of P, to execute. If T(P) does not deadlock or
>produce "false", P is type-safe.

This aproach is taken by Penn State's active network's program for the PLAN 
language (which has its roots in ML). I have not seen it applied to Java by 
anyone. My group is considering it. A question I have concerns the expressive 
power a "loop free" language has to the developer. What can and cannot be 
described under such constraints? I would also be interested in what tradeoffs 
work well between load-time/verification time and expressability. How large 
can a classfile be?

>
>This approach differs slightly from the one that Sean and
>Javasoft have implemented. In their cases, there are separate
>*byte code verifiers*, usually largish (10-15 page) C programs,
>implementing a "data-flow" analysis, documented by 10-15 pages of
>informal English (I have not seen Sean's documentation yet),
>about whose correctness it is hard to say anything formal. 

I have not seen a formal desription of Java's bytecode verifier. I have 
emailed Sun and Princeton's SIP group- no reply.. anyone heard of efferts 
along these lines?

>In any case, the bottom line for this list should be, I think,
>that one can assume that the subproblem of verifying the
>type-safety of JVM-like bytecode is solved, with of course, one
>important caveat: native code. (A lot of the "core" classes in
>Java in fact contain natively implemented methods, and one has to
>use separate techniques for verifying them.) 

Dont download native code unless it is authenticated from someone you trust :-)
Software fault isolation is another avenue to investigate.

>I would not be too surprised if this approach (Java's approach)
>of doing "link-time" (i.e. first-instruction-execution-time)
>type-checking actually resulted in significant performance
>*improvements* over standard OS's. One thing to keep in mind is
>that Java's architecture is such that the bytecode verifier can
>in fact accept some input code *even if it is not type-safe in
>isolation* --- as long as it is type-safe when combined with the
>other classes that are already loaded. That is, there are some
>possibilities of doing inter-classfile optimizations (e.g.
>eliminating some classcasts generated by the compiler) in a
>typesafe way, that can produce even more performance benefits
>than are posible if one does "straight" Java verification, as
>done currently by Sun's Java systems.

How big a win would this be?

From kaffe@kaffe.org  Mon Oct 13 19:47:22 1997
From: kaffe@kaffe.org (David HM Spector)
Date: Mon, 13 Oct 1997 14:47:22 -0400
Subject: JavaOS
In-Reply-To: Your message of "Mon, 13 Oct 1997 19:24:11 +0200."
 <199710131724.TAA02484@diogenes.iit.upco.es>
Message-ID: <199710131847.OAA28502@wintermute.ny.zeitgeist.com>


From the OSKit web page:

The Current Release 

	The last release of the OS Toolkit is currently unavailable pending
	the preparation of a much enhanced, somewhat incompatible version due
	in June (Oh oh, we're into July already!). Contact Jay Lepreau if you
	have an urgent need for the previous release (but don't hold your
	breath).


I am interested in evaulating OSKit for a commercial project and
wanted to find out a ball-park for licensing fees; I had mailed and
called a number of times and got no response.  Not very encouraging.

Last week on the OSKit-user list the project coordinator Jay Lepreau
stated that they would try to put up a current snaphot after the
SOPS16 conference.

OSKit looks like a very kool piece of work, but the caveat from the
authors themselves seems to be a good idea.

As for the non-commcercial use clause, it would be interesting to see
what the real story actually is.  The OSKit docs clearly state that
their work is derived from other GPLd sources including
Linux... (note: not an opinion... just an observation...)

David

-- 
-------------------------------------------------------------------------------
David HM Spector                                         spector@zeitgeist.com
Network Design & Infrastructure Security                 voice: +1 212.579.8573
Amateur Radio: W2DHM (ex-N2BCA) (ARRL life member)       GridSquare: FN30AS
-.-. --- -. -. . -.-. -  .-- .. - ....  .- -- .- - . ..- .-.  .-. .- -.. .. ---
"New and stirring things are belittled because if they are not belittled, 
the humiliating question arises, 'Why then are you not taking part in them?'"
                                                        --H. G. Wells

From kaffe@kaffe.org  Mon Oct 13 20:55:13 1997
From: kaffe@kaffe.org (James A. Hillyerd)
Date: Mon, 13 Oct 1997 12:55:13 -0700 (PDT)
Subject: Utah OSKit/JavaOS
Message-ID: <Pine.LNX.3.95.971013125028.17003A-100000@animal.blarg.net>

Here is a link to a page that talks a little about Utah's Java Computer:

http://www.cs.utah.edu/projects/flux/oskit/javaos.html

It says it will be free for non-commercial use, and that a full release
was due out May 97.  I see that they are behind on almost all of their
deadlines.  (Must be a Microsoft train ground.)

I am curious, does Kaffe's license allow someone to port it, then sell it
for commercial use?  That doesn't seem right to me.  [Perhaps the JavaOS
just plain isn't available for commercial use, since they don't explicitly
say they are selling it.]

-james

[  James A. Hillyerd (JH2162) - james@blarg.net - Web Developer  ]
[  http://www.blarg.net/~james/   http://www.hyperglyphics.com/  ]
[ 1024/B11C3751 CA 1C B3 A9 07 2F 57 C9  91 F4 73 F2 19 A4 C5 88 ]

From kaffe@kaffe.org  Mon Oct 13 20:52:48 1997
From: kaffe@kaffe.org (Pramod B. S.)
Date: Tue, 14 Oct 1997 01:22:48 +0530 (IST)
Subject: JavaOS
In-Reply-To: <19971013180200.04495@pcnov032.win.tue.nl> from "Mark Huizer" at Oct 13, 97 06:02:00 pm
Message-ID: <9710131952.AA11370@rishi.serc.iisc.ernet.in>

> Can you give me some URL or mailaddress so I can look at it? Sounds
> cool, and yes, I'm not commercial planning to use it :-)
> 
> Mark
> 
> 

http://www.cs.utah.edu/projects/flux/
look under the section titled Software Distribution Pending.

Pramod

From kaffe@kaffe.org  Tue Oct 14 02:21:46 1997
From: kaffe@kaffe.org (Neil A. Carson)
Date: Mon, 13 Oct 1997 18:21:46 -0700
Subject: Utah OSKit/JavaOS
References: <Pine.LNX.3.95.971013125028.17003A-100000@animal.blarg.net>
Message-ID: <3442C92A.2781E494@causality.com>

James A. Hillyerd wrote:

> I am curious, does Kaffe's license allow someone to port it, then sell it
> for commercial use?  That doesn't seem right to me.  [Perhaps the JavaOS
> just plain isn't available for commercial use, since they don't explicitly
> say they are selling it.]

You'd be surprised at the number of packages that are written like this:
For example, the whole of the NetBSD OS, and anything else under the
Berkeley licence.

	Cheers,

	Neil

-- 
Neil A. Carson                         Marketing Director
E-mail: neil@causality.com             Causality Limited (London, UK)
Mobile: +44 (0)370 593183              Tel/Fax: +44 (0)181 930 7408
Sent from a 233MHz StrongARM NC!       Web: http://www.causality.com.

From kaffe@kaffe.org  Tue Oct 14 02:28:11 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Mon, 13 Oct 1997 19:28:11 -0600 (MDT)
Subject: Utah OSKit/JavaOS
In-Reply-To: <Pine.LNX.3.95.971013125028.17003A-100000@animal.blarg.net> from "James A. Hillyerd" at Oct 13, 97 12:55:13 pm
Message-ID: <199710140128.TAA09531@sal.cs.utah.edu>


 Hi,

I'm the guy from the University of Utah who did the Kaffe to OS Kit
port. Unfortunately, I can't speak for my employer (Jay Lepreau), but
the situation as I see it is as follows.

We're behind our release schedule mainly for reasons of licensing which 
had to be resolved with our funders and the university. I cannot be more
specific at this point. Jay has told us that we will have at least a 
snapshot out by the end of next week. The license will allow free use 
for (at least) non-commercial purposes. The GNU-related/derived 
parts will obviously be distributed under the GNU license. 

As for the Kaffe port, it was done using an older version of the OS Kit
and using an older version of Kaffe (0.8.1). Although I don't currently
work on JavaOS, I want to make the port publicly available. As I'm doing
work-for-hire for the university, it won't be up to me to decide under
which license this will be distributed, but it will certainly not be in 
disagreement with Kaffe's license - which is a BSD-style license.
Before I make the port available, I would like to update to 0.9.2.

It is our hope that the OS Kit and the Kaffe port will spark research
in the areas that were discussed earlier in this thread. 

	- Godmar (gback@cs.utah.edu)

From kaffe@kaffe.org  Tue Oct 14 09:00:43 1997
From: kaffe@kaffe.org (Luis L. Fernandez)
Date: Tue, 14 Oct 1997 10:00:43 +0200
Subject: JavaOS
In-Reply-To: <344257A7.1B4@switch.rockwell.com>
References: <199710130657.IAA00386@diogenes.iit.upco.es>
 <344257A7.1B4@switch.rockwell.com>
Message-ID: <199710140800.KAA04485@diogenes.iit.upco.es>

Bartlee A. Anderson writes:
 > Luis,
 > I saw a reference to OSkit 5 months ago and its still not available
 > at the site you posted. What's the deal?

You are right, but this site give you a link to the person who is
responsible for OSKit project (Jay Lepreau). If you need the last
version of this library, please contact with him
(lepreau@cs.uath.edu).

Anyway if you have an urgent need I'm going to leave the oskit-0.60
release, the last one I have used, in the site:

ftp://nautilus.gsyc.inf.uc3m.es/pub/

Regards from Spain, Bart. If you need something else please, let me
know!!.

	Luis.

P.S: We have used OSKit to develop a "new" minimal micro-kernel from
the scratch. See "http://www.iit.upco.es/~luisl/advice.html" for more
info about this one. We have reduced the development time for six
months by using this toolkit. In my point, this one is a great effort
to lighten the OS development task.


o-------------------------------o------------------------------------o
| Luis Lopez Fernandez          |  e-mail: luisl@iit.upco.es         |   
| Industrial Systems Division   |          llf@computer.org          |
|                               |  http://www.iit.upco.es/~luisl     |      
o-------------------------------o------------------------------------o
|              Technological Research Institute                      |
|          Santa Cruz de Marcenado, 26 - Planta 1 -                  |
|	            28015 Madrid - SPAIN                             |
|      Tel: 34-1-5242800 (ext. 2706) - FAX: 34-1-5423176             |
o--------------------------------------------------------------------o

From kaffe@kaffe.org  Tue Oct 14 16:44:59 1997
From: kaffe@kaffe.org (Peter Michael Bertelsen)
Date: Tue, 14 Oct 1997 17:44:59 +0200
Subject: Formal JVM specification
In-Reply-To: <199710131742.NAA00507@postman.opengroup.org> (message from Dan
 Lambright on Mon, 13 Oct 1997 13:42:49 -0400)
Message-ID: <199710141544.RAA10521@ellemose.dina.kvl.dk>

Dan Lambright wrote:
> I have not seen a formal desription of Java's bytecode verifier. I have 
> emailed Sun and Princeton's SIP group- no reply.. anyone heard of efferts 
> along these lines?

I've written a report, `Semantics of Java Byte Code', giving a formal
description of the JVM and the bytecode instructions.  The JVM is
modelled as an infinite state machine, and each byte code instruction
is described by a transition rule, describing the state change and the
conditions under which execution succeeds.  The report also lists a
number of errors and ambiguities in Sun's specification of the JVM.

This might be of interest to you, although some aspects of class file
verification are not covered, e.g. checking that the file has the
correct structure.  The report is available at:

  ftp://ftp.dina.kvl.dk/pub/Staff/Peter.Bertelsen/jvm-semantics.ps.gz

Also check out the `Defensive JVM' project at Computational Logic:

  http://www.cli.com/software/djvm/

-Peter Bertelsen

From kaffe@kaffe.org  Tue Oct 14 18:00:21 1997
From: kaffe@kaffe.org (Bartlee A. Anderson)
Date: Tue, 14 Oct 1997 12:00:21 -0500
Subject: JavaOS
References: <199710130657.IAA00386@diogenes.iit.upco.es>
 <344257A7.1B4@switch.rockwell.com> <199710140800.KAA04485@diogenes.iit.upco.es>
Message-ID: <3443A525.7EE6@switch.rockwell.com>

Getting it now,
Thanks,
	Bart

Luis L. Fernandez wrote:
> 
> Bartlee A. Anderson writes:
>  > Luis,
>  > I saw a reference to OSkit 5 months ago and its still not available
>  > at the site you posted. What's the deal?
> 
> You are right, but this site give you a link to the person who is
> responsible for OSKit project (Jay Lepreau). If you need the last
> version of this library, please contact with him
> (lepreau@cs.uath.edu).
> 
> Anyway if you have an urgent need I'm going to leave the oskit-0.60
> release, the last one I have used, in the site:
> 
> ftp://nautilus.gsyc.inf.uc3m.es/pub/
> 
> Regards from Spain, Bart. If you need something else please, let me
> know!!.
> 
>         Luis.
> 
> P.S: We have used OSKit to develop a "new" minimal micro-kernel from
> the scratch. See "http://www.iit.upco.es/~luisl/advice.html" for more
> info about this one. We have reduced the development time for six
> months by using this toolkit. In my point, this one is a great effort
> to lighten the OS development task.
> 
> o-------------------------------o------------------------------------o
> | Luis Lopez Fernandez          |  e-mail: luisl@iit.upco.es         |
> | Industrial Systems Division   |          llf@computer.org          |
> |                               |  http://www.iit.upco.es/~luisl     |
> o-------------------------------o------------------------------------o
> |              Technological Research Institute                      |
> |          Santa Cruz de Marcenado, 26 - Planta 1 -                  |
> |                   28015 Madrid - SPAIN                             |
> |      Tel: 34-1-5242800 (ext. 2706) - FAX: 34-1-5423176             |
> o--------------------------------------------------------------------o

-- 
Bartlee A. Anderson     |  Sys. Test (ISDN-Commands-Trans Link-Tools)
Rockwell International  |                Electronic Commerce Division
300 Bauman Ct.          |                    banders@ecd.rockwell.com
MS 933-605      Opinions my own, not Rockwell's  VOICE (630) 227-8975
Wood Dale, IL 60191     |                        FAX   (630) 227-9771

From kaffe@kaffe.org  Wed Oct 15 10:17:22 1997
From: kaffe@kaffe.org (Malek Shabou)
Date: Wed, 15 Oct 1997 10:17:22 +0100
Subject: Pb with kaffe on win95
Message-ID: <199710150917.KAA29987@saturne.inria.fr>

 Hello
 I've compiled kaffe on a win95 succefuly, but when i try to run HelloWorldApp an error message apers:

kaffe -v -verbosejit HelloWorldApp

Loading class 'java/lang/Throwable'.
Loading class 'java/lang/IllegalArgumentException'.
Loading class 'java/lang/RuntimeException'.
Loading class 'java/lang/IllegalThreadStateException'.
Loading class 'java/lang/InterruptedException'.
Loading class 'java/lang/NoSuchMethodError'.
Loading class 'java/lang/IncompatibleClassChangeError'.
Loading class 'java/lang/LinkageError'.
Loading class 'java/lang/Error'.
Loading class 'java/lang/SecurityManager'.
Loading class 'java/lang/StringBuffer'.
Loading class 'java/lang/ThreadDeath'.
Loading class 'java/lang/ThreadGroup'.
Loading class 'java/lang/CloneNotSupportedException'.
Loading class 'java/lang/Integer'.
Loading class 'java/lang/Number'.
Loading class 'java/io/PrintStream'.
Loading class 'java/io/FilterOutputStream'.
Loading class 'java/io/OutputStream'.
Loading class 'java/lang/NullPointerException'.
Loading class 'sun/misc/VM'.
<JIT: java/lang/ThreadGroup.add(Ljava/lang/Thread;)V time 0ms (0ms) @ 0x4b82818>


(C:\KAFFE\BIN\KAFFE.EXE 1000) In cygwin_except_handler



  can any one help me?


A+


-- 
Malek Shabou				E-mail: Malek.Shabou@inria.fr
NIC France
INRIA-Rocquencourt Domaine de Voluceau
B.P.105 78153 Le Chesnay Cedex

From kaffe@kaffe.org  Wed Oct 15 16:52:09 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Thu, 16 Oct 1997 01:52:09 +1000 (EST)
Subject: NoSuchMethodError
Message-ID: <199710151552.BAA12624@zen.quick.com.au>

I was trying to get Servlets running under appache on a SunOS system
today using kaffe as the JVM.  The same apache config etc on a Solaris
system using jdk1.1.4 worked fine so I know everything other than
kaffe is cool (I tried 0.8.4,0.9.1,0.9.2).

Anyway, whenever I try to run a servlet via the jserv_mod[ule] kaffe
just says:

NoSuchMethodError

which is obviously sub-optimal from a debugging point of view.

I've not looked myself, but would it be difficult to modify things so
that one learned which method of which class was not found?

I've not yet tried using other servlet engines with kaffe - has anyone
had success with kaffe running servlets?

--sjg

From kaffe@kaffe.org  Wed Oct 15 21:48:28 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Wed, 15 Oct 1997 14:48:28 -0600 (MDT)
Subject: NoSuchMethodError
In-Reply-To: <199710151552.BAA12624@zen.quick.com.au> from "Simon J. Gerraty" at Oct 16, 97 01:52:09 am
Message-ID: <199710152048.OAA24151@lal.cs.utah.edu>

'Simon J. Gerraty' wrote:
> I've not looked myself, but would it be difficult to modify things so
> that one learned which method of which class was not found?

I've made the changes myself, but my local copy of Kaffe is whacked in
a number of respects, and you don't want to see my diff.  The actual
changes are pretty simple.  In kaffe/kaffevm/, grep for
'NoSuchMethodError', in many of the cases (e.g., in support.c) you'll
want to change those instances to 'NoSuchMethodErrorMsg(method_name)',
where "method_name" is some C string that describes the method.
For a simple hack, just use the method_name passed into the function,
in other places you may want to construct a more informative message
(e.g., "no such static method", or something.)  

Now you just need to define NoSuchMethodErrorMsg(M), which you'll put
in errors.h:
#define NoSuchMethodErrorMsg(M)  NEW_LANG_EXCEPTION_MESSAGE(NoSuchMethodError, M)

Hope this helps more than a diff....

-Pat


From kaffe@kaffe.org  Thu Oct 16 04:51:01 1997
From: kaffe@kaffe.org (Michael Madore)
Date: Wed, 15 Oct 1997 20:51:01 -0700
Subject: no math in shared library path
Message-ID: <3.0.1.32.19971015205101.006aaf88@sunshine>

Hi,

I'm running Kaffe 0.9.2 on Debian Linux (2.0.30 kernel).  When I attempt to
use the BigInteger class in my program I get the following error:

   Java.lang.UnsatisfiedLinkError:   no math in shared library path

I'm using Oracle's thin JDBC drivers, which apparently make use of this
class.  The error occurs if I use the class in a simple program without
JDBC also.  If I run the compiled program on an NT machine, I don't have
any problems.  Is this caused by my Linux installation, or by Kaffe?

TIA

Mike Madore

From kaffe@kaffe.org  Thu Oct 16 09:04:36 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Thu, 16 Oct 1997 18:04:36 +1000 (EST)
Subject: NoSuchMethodError
References: <199710152048.OAA24151@lal.cs.utah.edu>
Message-ID: <199710160804.SAA16176@gate.quick.com.au>

>'Simon J. Gerraty' wrote:
>> I've not looked myself, but would it be difficult to modify things so
>> that one learned which method of which class was not found?

>I've made the changes myself, but my local copy of Kaffe is whacked in

Thanks, I've made the appropriate changes and re-installed kaffe
I could not of course get it ti generate that error... :-)
Might have been kaffe-0.8.4 doing that :-(, anyway the patches
seem a neat addition - though I'll need to test what the result
looks like.

The good news is that after getting no where via jserv, I ran:

java sun.servlet.http.HttpServer

with appropriate CLASSPATH (java is my java_wrapper which runs kaffe) 
and using lynx, I can get SnoopServlet running.  
So presumably the problem is related to jserv.
I'll try soem other servlet engines and see how it goes.

--sjg




-- 
Simon J. Gerraty        <sjg@quick.com.au>

#include <disclaimer>   /* imagine something _very_ witty here */

From kaffe@kaffe.org  Thu Oct 16 10:13:38 1997
From: kaffe@kaffe.org (Malek Shabou)
Date: Thu, 16 Oct 1997 10:13:38 +0100
Subject: kaffe 0.9.2 with win95
Message-ID: <199710160913.KAA04563@saturne.inria.fr>

 Hello,
 i've compiled kaffe 0.9.2 under win95 envirenment, and when i trie to execute 
HelloWorldApp or any other application i get this:
Loading class 'java/lang/Throwable'.
Loading class 'java/lang/IllegalArgumentException'.
Loading class 'java/lang/RuntimeException'.
Loading class 'java/lang/IllegalThreadStateException'.
Loading class 'java/lang/InterruptedException'.
Loading class 'java/lang/NoSuchMethodError'.
Loading class 'java/lang/IncompatibleClassChangeError'.
Loading class 'java/lang/LinkageError'.
Loading class 'java/lang/Error'.
Loading class 'java/lang/SecurityManager'.
Loading class 'java/lang/StringBuffer'.
Loading class 'java/lang/ThreadDeath'.
Loading class 'java/lang/ThreadGroup'.
Loading class 'java/lang/CloneNotSupportedException'.
Loading class 'java/lang/Integer'.
Loading class 'java/lang/Number'.
Loading class 'java/io/PrintStream'.
Loading class 'java/io/FilterOutputStream'.
Loading class 'java/io/OutputStream'.
Loading class 'java/lang/NullPointerException'.
Loading class 'sun/misc/VM'.
<JIT: java/lang/ThreadGroup.add(Ljava/lang/Thread;)V time 0ms (0ms) @ 0x4b82818>


(C:\KAFFE\BIN\KAFFE.EXE 1000) In cygwin_except_handler





can any help me,

Cordialement,

-- 
Malek Shabou				E-mail: Malek.Shabou@inria.fr
NIC France
INRIA-Rocquencourt Domaine de Voluceau
B.P.105 78153 Le Chesnay Cedex


From kaffe@kaffe.org  Thu Oct 16 18:46:21 1997
From: kaffe@kaffe.org (James Cooper)
Date: Thu, 16 Oct 1997 10:46:21 -0700 (PDT)
Subject: solaris 2.5 + socket = core dump
Message-ID: <Pine.GSO.3.95.971016104047.19310B-100000@caustik.organic.com>


Hi,

I'm trying to get Apache's JServ running under kaffe
(http://java.apache.org).  JServ is a runtime environment for servlets
that work with apache.  It doesn't use any native methods of its own.

A little info about my setup:
  Sparc 5 - Solaris 2.5
  kaffe 0.9.1
  classes.zip from the Sun JDK 1.1.2

HelloWorldApp runs fine, as does the javac wrapper, so I was psyched.  I
configured JServ to use the kaffe VM instead of the Sun VM, and fired it
up.

JServ starts up fine, and begins listening on a socket.  However, when an
incoming connection comes in it dumps core.  Here's the stack trace from
gdb:

(gdb) where
#0  0xef6f4140 in _kill ()
#1  0xef6b965c in abort ()
#2  0xef4a2a1c in java_net_PlainSocketImpl_socketGetOption (this=0x276490, 
    val=15) at ./java.net/PlainSocketImpl.c:242
#3  0x29f140 in _end ()
#4  0x2a32c4 in _end ()
#5  0x2a7164 in _end ()
#6  0x271cc4 in _end ()
#7  0xef754420 in do_execute_java_method (ee=0xef774528, obj=0x14f058, 
    method_name=0x14f058 "", signature=0xef774528 "()V", mb=0x130890, 
    isStaticCall=0) at support.c:83
#8  0xef751264 in firstStartThread () at thread.c:206
#9  0xef752c30 in reschedule () at thread-internal.c:497
#10 0xc in ?? ()


I know very little about gdb, so that didn't mean much to me, but I
thought it might be helpful to someone here.

I've noticed that other folks on the list are successfully running servers
under kaffe, so I imagine it's something funky with my setup.  If anyone
has any suggestions on how to better diagnose the cause of this problem
I'd be appreciative.

Thanks

-- James

--------------------------------------------------------------------------------
                              James Paul Cooper
                      Organic Online - pixel@organic.com
                          ......Know Future......

From kaffe@kaffe.org  Fri Oct 17 05:44:26 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Fri, 17 Oct 1997 14:44:26 +1000 (EST)
Subject: kaffe can handle servlets
Message-ID: <199710170444.OAA29667@gate.quick.com.au>

FYI, a trivial mod to apache's mod_jserv.c (replace "rb+" with "r+b")
gets servlets running on SunOS 4.1.4 with kaffe-0.9.2.

When I try this on NetBSD with kaffe-0.9.1 I just get:

[Fri Oct 17 14:20:41 1997] access to /share/java/JSDK1.0.1/servlets/DateServlet failed for xx.xx.xx.xx, reason: Premature end of script headers

and of course I've not got 0.9.2 working yet.

--sjg


From kaffe@kaffe.org  Fri Oct 17 05:41:00 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Fri, 17 Oct 1997 14:41:00 +1000 (EST)
Subject: kaffe-0.9.2: core dumps on NetBSD/i386
Message-ID: <199710170441.OAA29603@gate.quick.com.au>


kaffe-0.9.2 builds ok on NetBSD/i386 but if I try and
run java foo which should produce 
java.lang.ClassNotFoundException: foo
I get
Illegal instruction (core dumped)

The stack trace looks a bit odd, so I tried:

sjg:202$ ulimit -s 32768
sjg:203$ java foo
[wait....,wait...,....]
/share: write failed, file system is full
Illegal instruction 

I would guess that something is recursing itself to death.

--sjg

From kaffe@kaffe.org  Fri Oct 17 05:25:16 1997
From: kaffe@kaffe.org (PookMan)
Date: Fri, 17 Oct 1997 00:25:16 -0400 (EDT)
Subject: Kaffe or coffee ?
Message-ID: <Pine.SUN.3.91.971017002432.8906B-100000@lisa.cs.mcgill.ca>


 I am sorry if this sounds like a stupid question but In the README file, 
we read this :

For a Kaffe system you need the following parts:

 kaffe-0.9.1.tgz                        Virtual machine and all bits
                                        provided by tjwassoc.co.uk
 kaffe-0.9.1-package-javasoft.com.tgz   Sun's class libraries.


If you want to use graphics you'll need one of the following:

 kaffe-0.9.1-package-biss-net.com.tgz   AWT graphics support.
 kaffe-0.9.1-package-epfl.ch.tgz        Alternative graphics support.


  I simply want to know where we can get those packages ?? 
At the corresponding site ? Where exactly ? ftp address ?

 THank you for your help

From kaffe@kaffe.org  Fri Oct 17 08:43:22 1997
From: kaffe@kaffe.org (PookMan)
Date: Fri, 17 Oct 1997 03:43:22 -0400 (EDT)
Subject: Kaffe or coffee ? (fwd)
Message-ID: <Pine.SUN.3.91.971017034047.16050A-100000@lisa.cs.mcgill.ca>


 Please forget my previous question and focus on this one :)
 
I am using BSDI 2.1 and I am trying to install Kaffe

 I get the following error at the compiling

 true --def ./libkaffe_bissawt.def --output-exp lib.exp --output-lib 
libkaffe_bissawt.a --dllname libkaffe_bissawt.a
touch lib.exp
ar cr libkaffe_bissawt.a.0.87 biss_awt_kernel_NativeLib.o  color.o  
display.o  font.o  graphics.o  image.o  window.o
ar: warning: biss_awt_kernel_NativeLib.o truncated to biss_awt_kernel
ranlib libkaffe_bissawt.a.0.87
rm -f libkaffe_bissawt.a
ln -s libkaffe_bissawt.a.0.87 libkaffe_bissawt.a
gcc -g -O -DNO_SHARED_LIBRARIES -I. -I. -I./../../config -I../../config 
-I../../include -I./../../include -DTRANSLATOR -I./jit -DKVER=\"0.91\" 
-c  ./jit/funcs.c
funcs.c:8805:Unknown pseudo-op:  '.global'
funcs.c:8805:Rest of line ignored. 1st junk character valued 95 (_).
*** Error code 1

Stop.
*** Error code 1

Stop.
*** Error code 1

Stop.
 
 Does anyone has any idea on how to solve this problem ?

 I would really appreciate any tip from you ...

 thank you

From kaffe@kaffe.org  Fri Oct 17 08:40:07 1997
From: kaffe@kaffe.org (Artur Biesiadowski)
Date: Fri, 17 Oct 1997 09:40:07 +0200 (MET DST)
Subject: Java inside gcc ?
Message-ID: <Pine.GSO.3.96L01AT.971016195031.2682A-100000@sunrise>



At what stage is project of integrating java/kaffe with gcc ?

Artur

From kaffe@kaffe.org  Fri Oct 17 11:50:16 1997
From: kaffe@kaffe.org (Fathi.DJEBBARI)
Date: Fri, 17 Oct 1997 11:50:16 +0100
Subject: Kaffe or coffee ? (fwd)
Message-ID: <199710171050.LAA02589@bcv64s3e.vz.cit.alcatel.fr>


pooky wrote 
> From kaffe-approval@soi.city.ac.uk Fri Oct 17 12:42 MET 1997
> X-Authentication-Warning: vega.soi.city.ac.uk: lp set sender to owner-kaffe@soi.city.ac.uk using -f
> Date: Fri, 17 Oct 1997 03:43:22 -0400 (EDT)
> From: PookMan <pooky@cs.mcgill.ca>
> To: kaffe@kaffe.org
> Subject: Kaffe or coffee ? (fwd)
> Mime-Version: 1.0
> Sender: owner-kaffe@soi.city.ac.uk
> Reply-To: kaffe@kaffe.org
> Content-Type> : > TEXT/PLAIN> ; > charset=US-ASCII> 
> Content-Length: 1056
> 
> 
>  Please forget my previous question and focus on this one :)
>  
> I am using BSDI 2.1 and I am trying to install Kaffe
> 
>  I get the following error at the compiling
> 
>  true --def ./libkaffe_bissawt.def --output-exp lib.exp --output-lib 
> libkaffe_bissawt.a --dllname libkaffe_bissawt.a
> touch lib.exp
> ar cr libkaffe_bissawt.a.0.87 biss_awt_kernel_NativeLib.o  color.o  
> display.o  font.o  graphics.o  image.o  window.o
> ar: warning: biss_awt_kernel_NativeLib.o truncated to biss_awt_kernel
> ranlib libkaffe_bissawt.a.0.87
> rm -f libkaffe_bissawt.a
> ln -s libkaffe_bissawt.a.0.87 libkaffe_bissawt.a
> gcc -g -O -DNO_SHARED_LIBRARIES -I. -I. -I./../../config -I../../config 
> -I../../include -I./../../include -DTRANSLATOR -I./jit -DKVER=\"0.91\" 
> -c  ./jit/funcs.c
> funcs.c:8805:Unknown pseudo-op:  '.global'
> funcs.c:8805:Rest of line ignored. 1st junk character valued 95 (_).
> *** Error code 1
> 
> Stop.
> *** Error code 1
> 
> Stop.
> *** Error code 1
> 
> Stop.
>  
>  Does anyone has any idea on how to solve this problem ?
> 
>  I would really appreciate any tip from you ...
> 
>  thank you
> 
> 
I am not really sure of my answer, but it seems to me that bsdi assembler
work with globl lalel instead of global.
Bye
Fathi

From kaffe@kaffe.org  Fri Oct 17 13:54:02 1997
From: kaffe@kaffe.org (Martin Lanzen)
Date: Fri, 17 Oct 1997 14:54:02 +0200 (MET DST)
Subject: Kaffe or coffee ?
In-Reply-To: <Pine.SUN.3.91.971017002432.8906B-100000@lisa.cs.mcgill.ca>
Message-ID: <199710171254.OAA27860@puh>

On 17 Okt , PookMan wrote:
> 
>  I am sorry if this sounds like a stupid question but In the README file, 
> we read this :
> 
> For a Kaffe system you need the following parts:
> 
>  kaffe-0.9.1.tgz                        Virtual machine and all bits
>                                         provided by tjwassoc.co.uk
>  kaffe-0.9.1-package-javasoft.com.tgz   Sun's class libraries.
> 
> 
> If you want to use graphics you'll need one of the following:
> 
>  kaffe-0.9.1-package-biss-net.com.tgz   AWT graphics support.
>  kaffe-0.9.1-package-epfl.ch.tgz        Alternative graphics support.
> 
> 
>   I simply want to know where we can get those packages ?? 
> At the corresponding site ? Where exactly ? ftp address ?
> 
>  THank you for your help
> 
I found these files at:
 
http://www.oasis.leo.org/java/machines/kaffe/00-index.html
-- 
                                                           
/Martin                                        

From kaffe@kaffe.org  Fri Oct 17 18:46:19 1997
From: kaffe@kaffe.org (Vik Sohal)
Date: Fri, 17 Oct 1997 10:46:19 -0700
Subject: Location of packages for kaffe
Message-ID: <2.2.32.19971017174619.00bc7a8c@wizard.lynx.com>


You can get the BISS file from:

http://btp1da.phy.uni-bayreuth.de/~werner/FreeBSD/fbsd_pub_ftp/distfiles/

The epfl file can be gotten from:

http://gd.tuwien.ac.at/languages/java/kaffe/ghindex.shtml

Vik
***************************************************************
* Vik Sohal                     *  email: vik@lynx.com        *
* Lynx Real-Time Systems        *  voice: (408) 879-3900 x119 *
* 2239 Samaritan Drive          *    fax: (408) 879-3920      *
* San Jose, CA 95124            *                             *
* ------------------------------*-----------------------------*
* cute msg: "We just turned it on, and it's I.Q. is already   *
*             in the thousands!"                              *
***************************************************************

From kaffe@kaffe.org  Fri Oct 17 20:42:20 1997
From: kaffe@kaffe.org (Per Bothner)
Date: Fri, 17 Oct 1997 12:42:20 -0700
Subject: Java inside gcc ?
In-Reply-To: Your message of "Fri, 17 Oct 1997 09:40:07 +0200."
 <Pine.GSO.3.96L01AT.971016195031.2682A-100000@sunrise>
Message-ID: <199710171941.MAA29724@cygnus.com>

> At what stage is project of integrating java/kaffe with gcc ?

Assuming you mean:  Integrating gcc-compiled Java code with Kaffe,
then:  Fairly far.  The major outstanding issue is exception handling.

	--Per Bothner
Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner

From kaffe@kaffe.org  Fri Oct 17 23:55:20 1997
From: kaffe@kaffe.org (Per Bothner)
Date: Fri, 17 Oct 1997 15:55:20 -0700
Subject: Java inside gcc ?
In-Reply-To: Your message of "Fri, 17 Oct 1997 17:20:43 CDT."
 <199710172220.RAA16096@compound.east.sun.com>
Message-ID: <199710172254.PAA08199@cygnus.com>

Tony Kimball <alk@pobox.com> writes:

> Will it be part of egcs?

That is one possibility, but it has not been decided.
(It partly depends on the egcs release plans.)

	--Per Bothner
Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner

From kaffe@kaffe.org  Sat Oct 18 02:28:10 1997
From: kaffe@kaffe.org (David Morton)
Date: Fri, 17 Oct 1997 20:28:10 -0500 (CDT)
Subject: Java inside gcc ?
In-Reply-To: <199710171941.MAA29724@cygnus.com>
Message-ID: <XFMail.971017202810.mortonda@dgm.cadvantage.com>

I'm really getting anxious... :)

On 17-Oct-97 Per Bothner wrote:
>> At what stage is project of integrating java/kaffe with gcc ?
> 
> Assuming you mean:  Integrating gcc-compiled Java code with Kaffe,
> then:  Fairly far.  The major outstanding issue is exception handling.
> 
>       --Per Bothner
> Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner

--------------------------------------------------------
David Morton         See my webpage for my public pgp key                 
E-Mail: mortonda@esuvm.emporia.edu           
WWW: http://thinker.emporia.edu/~mortonda/home.html
Date: 17-Oct-97 Time: 20:27:39  CST -600 GMT        

From kaffe@kaffe.org  Sat Oct 18 11:35:46 1997
From: kaffe@kaffe.org (Preben Randhol)
Date: 18 Oct 1997 12:35:46 +0200
Subject: Graphics
Message-ID: <jclk9fbmk8t.fsf@iq.pvv.ntnu.no>

I've tried to run a small java-program in X Windows. It works with the
ordinary java interpreter, but when I try to run it with kaffe there is just
a lot of error-msgs. It is only a small window with a button on it. Nothing
fancy. Kaffe works great with programs that only outputs text to STDOUT
though. Is there some thing I need ? I have installed the biss and eplch packages.

Thanks in advance

:info
Kaffe 0.9.1
Platform i586 with RedHat 4.2 Linux
Kernel 2.0.30

-- 
Preben Randhol                    | Soldiers are followers to the
Tlf:   73940929    Arb: 73594076  | extreme.
Epost: randhol@pvv.org            |
http://www.pvv.org/~randhol/      |     

From kaffe@kaffe.org  Sat Oct 18 21:22:57 1997
From: kaffe@kaffe.org (Corey Minyard)
Date: 18 Oct 1997 15:22:57 -0500
Subject: Java inside gcc ?
In-Reply-To: Per Bothner's message of Fri, 17 Oct 1997 12:42:20 -0700
References: <199710171941.MAA29724@cygnus.com>
Message-ID: <m2iuuuzuqm.fsf@wf-rch.cirr.com>

Per Bothner <bothner@cygnus.com> writes:

> 
> > At what stage is project of integrating java/kaffe with gcc ?
> 
> Assuming you mean:  Integrating gcc-compiled Java code with Kaffe,
> then:  Fairly far.  The major outstanding issue is exception handling.
> 
> 	--Per Bothner
> Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner
> 

Just a thought, you might want to talk to the GNAT people (the GNU
version of Ada supported by Ada Core Technologies, if you don't know)
about exception handling.  Ada and Java exception handling are pretty
close semantically and they might have useful stuff for you.

-- 
Corey Minyard               Internet:  minyard@acm.org
  Work: minyard@nortel.ca       UUCP:  minyard@wf-rch.cirr.com

From kaffe@kaffe.org  Mon Oct 20 07:54:35 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Mon, 20 Oct 1997 16:54:35 +1000 (EST)
Subject: kaffe can handle servlets
References: <199710170444.OAA29667@gate.quick.com.au>
Message-ID: <199710200654.QAA05675@gate.quick.com.au>

>FYI, a trivial mod to apache's mod_jserv.c (replace "rb+" with "r+b")
>gets servlets running on SunOS 4.1.4 with kaffe-0.9.2.

>When I try this on NetBSD with kaffe-0.9.1 I just get:

>[Fri Oct 17 14:20:41 1997] access to /share/java/JSDK1.0.1/servlets/DateServlet failed for xx.xx.xx.xx, reason: Premature end of script headers

The solution to this was the following patch:

*** kaffe-0.9.1/packages/tjwassoc.co.uk/APInet/lib/java.net/PlainSocketImpl.c.~1~    Fri Jun 27 01:02:10 1997
--- kaffe-0.9.1/packages/tjwassoc.co.uk/APInet/lib/java.net/PlainSocketImpl.c        Sun Jul 27 10:26:50 1997
***************
*** 233,243 ****
  void
  java_net_PlainSocketImpl_socketSetOption(struct Hjava_net_PlainSocketImpl* this, jint v1, jint /* bool */ v2, struct Hjava_lang_Object* v3)
  {
!       abort();
  }
  
  jint
  java_net_PlainSocketImpl_socketGetOption(struct Hjava_net_PlainSocketImpl* this, jint val)
  {
!       abort();
  }
--- 233,245 ----
  void
  java_net_PlainSocketImpl_socketSetOption(struct Hjava_net_PlainSocketImpl* this, jint v1, jint /* bool */ v2, struct Hjava_lang_Object* v3)
  {
!       /* abort(); /* no thanks --sjg */
  }
  
  jint
  java_net_PlainSocketImpl_socketGetOption(struct Hjava_net_PlainSocketImpl* this, jint val)
  {
!       /* abort(); /* no thanks --sjg */
!       val = 0;
!       return 0;
  }

-- 
Simon J. Gerraty        <sjg@quick.com.au>

#include <disclaimer>   /* imagine something _very_ witty here */

From kaffe@kaffe.org  Mon Oct 20 10:52:59 1997
From: kaffe@kaffe.org (Iso-H)
Date: Mon, 20 Oct 1997 12:52:59 +0300 (EET DST)
Subject: Solaris 2.5 + kaffe = core dump
In-Reply-To: <Pine.GSO.3.95.971016104047.19310B-100000@caustik.organic.com>
Message-ID: <Pine.LNX.3.95q.971020124525.9063A-100000@linray.ts.ray.fi>

On Thu, 16 Oct 1997, James Cooper wrote:

> 
> Hi,
> 
> I'm trying to get Apache's JServ running under kaffe
> (http://java.apache.org).  JServ is a runtime environment for servlets
> that work with apache.  It doesn't use any native methods of its own.
> 
> A little info about my setup:
>   Sparc 5 - Solaris 2.5
>   kaffe 0.9.1
>   classes.zip from the Sun JDK 1.1.2
> 

  Did you compile kaffe as dynamic? I can't get kaffe to work
  (under Solaris) if it is compiled as dynamic; it just dumps 
  core all the time!

  But if I use configure's option "--enable-staticlib" then kaffe
  works, but "KaffeVerifyBug" test gives this:

  java.lang.UnsatisfiedLinkError: no net in shared library path
        at java/lang/Throwable.<init>(line unknown, pc 16bf2c) 
        at java/lang/Error.<init>(line unknown, pc 18b654) 
        at java/lang/LinkageError.<init>(line unknown, pc 18b5dc) 
        at java/lang/UnsatisfiedLinkError.<init>(line unknown, pc 18b564) 
        at java/lang/Runtime.loadLibrary(line unknown, pc 16fbc0) 
        at java/lang/System.loadLibrary(line unknown, pc 18adf0) 
        at java/net/InetAddress.<clinit>(line unknown, pc 19219c) 
        at java/net/Socket.<init>(97)
        at KaffeVerifyBug.doit(21)
        at KaffeVerifyBug.main(9)


Iso-H

--

From kaffe@kaffe.org  Mon Oct 20 21:22:13 1997
From: kaffe@kaffe.org (Amitabh Arora)
Date: Mon, 20 Oct 1997 13:22:13 -0700
Subject: Binaries for Sun4.1 and HP 9.05 OSs
Message-ID: <01BCDD5B.2DFF26B0@amitabh@aspec.com>

Does anyone have Kaffe running on Sun 4.1.3 and HP-UX 9.05?
If yes, then please let me know how to download the binaries?

Thanks. -amitabh arora

From kaffe@kaffe.org  Mon Oct 20 23:10:10 1997
From: kaffe@kaffe.org (Alexandre Oliva)
Date: 20 Oct 1997 20:10:10 -0200
Subject: kaffe 0.9.* on IRIX 5.2: threads.s
Message-ID: <ord8l0gk71.fsf@grupiara.dcc.unicamp.br>

--Multipart_Mon_Oct_20_20:10:09_1997-1
Content-Type: text/plain; charset=US-ASCII

Hi there!

I've been discussing with Ian Lance Taylor, GNU binutils development
team leader, why I couldn't build kaffe 0.9.[012] on IRIX 5.2 if
threads.s is compiled with GNU as.  Compiling it with IRIX's built-in
as would do fine.

We couldn't decide whose fault it was exactly, but it is possible that
it is a bug in IRIX's ld, since the same object file, produced by GNU
as 2.8.1, is accepted by IRIX 5.3's ld.

Anyway, Ian has suggested the attached patch for threads.s.  It does
not affect the outcome at all, but it provides more accurate
information about the assembly code.

Regards,

-- 
Alexandre Oliva
mailto:oliva@dcc.unicamp.br mailto:aoliva@acm.org
http://www.dcc.unicamp.br/~oliva
Universidade Estadual de Campinas, SP, Brasil

--Multipart_Mon_Oct_20_20:10:09_1997-1
Content-Type: application/octet-stream; type=patch
Content-Disposition: attachment; filename="mipskaffe.diff"
Content-Transfer-Encoding: 7bit

--- config/mips/threads.s~	Fri Oct 10 20:38:39 1997
+++ config/mips/threads.s	Fri Oct 10 20:38:16 1997
@@ -11,7 +11,7 @@
  *
  */
 
-	.globl	_threadswitch
+	.globl	_threadswitch	.text
 _threadswitch:	
 	subu	$sp,$sp,32*4
 	sw	$2,  2*4($sp)
@@ -77,7 +77,7 @@
 	addu	$sp,$sp,32*4	
 	j	$31
 
-	.globl	_threadsp	
+	.globl	_threadsp	.text
 _threadsp:
 	move	$2,$sp
 	j	$31

--Multipart_Mon_Oct_20_20:10:09_1997-1--
From kaffe@kaffe.org  Tue Oct 21 03:33:13 1997
From: kaffe@kaffe.org (Michael Gesundheit)
Date: Mon, 20 Oct 1997 19:33:13 -0700 (PDT)
Subject: Kaffe bug on PC
Message-ID: <19971021023313.12883.rocketmail@send1.rocketmail.com>


Hi There,

I was able to build kaffe under VC++ 5.0. It does not
work yet but looks promissing. I have the following
problem:

In initialization time while executing initialization
of system class the vm gets into nested calls to
itself situation. In file kaffe.def:

define_insn(INVOKESTATIC)
{
......

#if defined(TRANSLATOR) && defined(HAVE_TRAMPOLINE))
  
call_indirect_const(&METHOD_NATIVECODE(method_method());
#else
   call(tmp);
#endif

........
}

In my configuration the call(tmp) is what compiled
and, if you look at this macro it is, basicly, calling
virtualMachine() which again gets back to INVOKESTATIC
case and like that - forever until the program runs
out of stack.

Any clue????

Thanks,

--Michael




_____________________________________________________________________
Sent by RocketMail. Get your free e-mail at http://www.rocketmail.com

From kaffe@kaffe.org  Tue Oct 21 18:18:27 1997
From: kaffe@kaffe.org (George Chung)
Date: Tue, 21 Oct 1997 10:18:27 -0700
Subject: Native threads and native methods on Solaris
Message-ID: <01BCDE0A.AC65E750.gchung@openhorizon.com>

Hi,
I'm new to this mailing list and couldn't really find anything in the 
archives, so here goes...

1) From looking at config/sparc/solaris2/*.c, it appears that the 
solaris/sparc implementation of Kaffe is based on a user level threads 
package. Is this correct?

2) If native methods are supported, either through JNI or through the old 
1.02 technique, then the native method library must be linked with this 
Kaffe user level threads package. Is this correct? Do you provide jacket 
routines for the synchronous system calls such as read, write, etc?

Regards,
______________________________________________________________________
George Chung
Open Horizon, Inc.
601 Gateway Blvd.
Suite 800
South San Francisco, CA 94080
t: (650) 869-2220
f: (650) 869-2201
e: gchung@openhorizon.com
w: http://www.openhorizon.com

Find out about Ambrosia(tm), the SECURE Publish/Subscribe Event
Management System for the Internet (and 100% Pure Java Certified),
at http://www.openhorizon.com.

From kaffe@kaffe.org  Tue Oct 21 19:10:33 1997
From: kaffe@kaffe.org (David Waite)
Date: Tue, 21 Oct 1997 14:10:33 -0400
Subject: ICQ for Java success?
Message-ID: <199710211839.OAA29478@m23.rickards.leon.k12.fl.us>

Has anyone had success in getting ICQ (http://www.mirabilis.com) to run
under kaffe? I've tried to get it started several times, there is a problem
with the released binaries in the linux jdk that prevents me from using them
with my system. I haven't tried kaffe yet though

ICQ is an internet pager concept: you add users to you rlist, then the
server tells you when they are on, and lets you send messages/ files and
chat with them.

-David Waite

From kaffe@kaffe.org  Wed Oct 22 08:28:50 1997
From: kaffe@kaffe.org (Nathan Gelbard)
Date: Wed, 22 Oct 1997 00:28:50 -0700
Subject: NetBSD-m68k-0.9.1 compile problem
Message-ID: <3.0.32.19971022002850.007dee70@engr.orst.edu>

I'm running NetBSD-m68k-hp300 1.2E

gcc -g -O  -I. -I. -I./../../config -I../../config -I../../include
-I./../../inc
lude -DINTERPRETER -I./intrp -DKVER=\"0.91\" -c -fpic ./intrp/machine.c
./intrp/machine.c: In function `virtualMachine':
./intrp/machine.c:141: structure has no member named `exception_table_len'
./intrp/machine.c:152: structure has no member named `ins'
./intrp/machine.c:178: structure has no member named `code'
./intrp/machine.c:182: structure has no member named `codelen'

I just grabbed the src tonight.

What to do?

	nate



[ Nathan Gelbard              ][ gelbard@engr.orst.edu           ] 
[ Oregon State University     ][ http://straylight.hurrah.com    ]
[ Student of Computer Science ][ Area Support Computer Assistant ]
From kaffe@kaffe.org  Wed Oct 22 19:03:49 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Wed, 22 Oct 1997 12:03:49 -0600 (MDT)
Subject: intrp/machine.c bug
Message-ID: <199710221803.MAA04478@lal.cs.utah.edu>

Hi,

I was encountering wacky behavior in my hacked copy of Kaffe when
running lots of threads in some simple tests.  (I couldn't get the
release version to duplicate the problem, but I think its still a
problem in 0.9.2.)  Basically, a bunch of stack variables were getting
trashed in intrp/machine.c::virtualMachine() (the interpreter loop.)
I ran 'gcc -O2 -Wall' on it, and gcc tells me that a bunch of
variables might get trashed by longjmp()---exactly the variables that
were getting trashed.  There is a single setjmp() in virtualMachine().
I got rid of the warnings (and the problem!) by putting the setjmp in
a separate function (I read that the problems with setjmp/longjmp are
limited to the active stack frame).  Attached is a patch for
intrp/machine.c.  Anyone with more experience with setjmp/longjmp care
to explain what was going on?  ;)

-Pat

--- /home/grad/tullmann/java/kaffe-0.9.2/kaffe/kaffevm/intrp/machine.c Sat Oct  4 04:26:57 1997
+++ /home/grad/tullmann/tmp/machine.c   Wed Oct 22 11:59:46 1997
@@ -68,6 +68,12 @@
 /* Misc stuff */
 Hjava_lang_Object* exceptionObject;
 
+static int
+exceptionSetjmp(vmException *mjbuf)
+{
+       return setjmp(mjbuf->jbuf);
+}
+
 void
 virtualMachine(methods* meth, slots* arg, slots* retval)
 {
@@ -140,7 +146,7 @@
                TCTX(currentThread)->exceptPtr = &mjbuf;
        }
        if (meth->exception_table != 0 || (methaccflags & ACC_SYNCHRONISED)) {
-               if (setjmp(mjbuf.jbuf) != 0) {
+               if (exceptionSetjmp(&mjbuf) != 0) {
                        assert(TCTX(currentThread)->exceptPtr == &mjbuf);
                        npc = mjbuf.pc;
                        sp = &lcl[meth->localsz + meth->stacksz - 1];

From kaffe@kaffe.org  Wed Oct 22 19:44:21 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Wed, 22 Oct 1997 12:44:21 -0600 (MDT)
Subject: bug in intrp/machine.c
Message-ID: <199710221844.MAA04631@lal.cs.utah.edu>

Doh!

I didn't test that solution quite as well as I should have!

setjmp/longjmp specify that you cannot return from a function that you
call setjmp in and expect it to work.  (So one problem goes away, but
Kaffe's exception handling ability is trashed.)  The standard solution
is to make all the questionable stack variables volatile...  I'll see
what that does...

-Pat



From kaffe@kaffe.org  Wed Oct 22 20:32:19 1997
From: kaffe@kaffe.org (Amitabh Arora)
Date: Wed, 22 Oct 1997 12:32:19 -0700
Subject: Binaries for Sun 4.1.3 and HP-UX 9.0 OS's
Message-ID: <01BCDEE6.8A8C0DD0@amitabh@aspec.com>

As the title says.  Does any body have the binaries for the above two
platforms.  It will save me a lot of work and I can spend more time writing
Java applications.  My compilations have failed in the past and I am unable
to devote time to fix the compiles.

Please please help.

Thanks. -amitabh arora
From kaffe@kaffe.org  Wed Oct 22 16:06:31 1997
From: kaffe@kaffe.org (Nikolai Vladychevski)
Date: Wed, 22 Oct 97 15:06:31 -0000
Subject: NetBSD-m68k-0.9.1 compile problem
Message-ID: <199710221957.UAA18996@home.virtual-pc.com>

>Subject:     NetBSD-m68k-0.9.1 compile problem
>Sent:        10/22/97 7:27
>Received:    10/22/97 9:33
>From:        Nathan Gelbard, gelbard@engr.orst.edu
>Reply-To:    kaffe@kaffe.org
>To:          kaffe@sarc.city.ac.uk
>
>I'm running NetBSD-m68k-hp300 1.2E
>
>gcc -g -O  -I. -I. -I./../../config -I../../config -I../../include
>-I./../../inc
>lude -DINTERPRETER -I./intrp -DKVER=\"0.91\" -c -fpic ./intrp/machine.c
>./intrp/machine.c: In function `virtualMachine':
>./intrp/machine.c:141: structure has no member named `exception_table_len'
>./intrp/machine.c:152: structure has no member named `ins'
>./intrp/machine.c:178: structure has no member named `code'
>./intrp/machine.c:182: structure has no member named `codelen'
>
>I just grabbed the src tonight.
>
>What to do?
>
>	nate
this question was answered in this list already, but anyway, i couldnt run
kaffe 9.1 on my Mac, due to problem with symbols of some libraries, this 
was
also discussed in this list, but when i compiled it, and ran, it told me 
about some
symbols problem. I didnt insist, but if you make it work, let me know i 
will fight again.

niko

From kaffe@kaffe.org  Sat Oct 25 02:08:49 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Sat, 25 Oct 1997 02:08:49 +0100
Subject: A Beautiful Painting!
Message-ID: <199710250108.CAA25266@cripplecock.sarc.city.ac.uk>

Dear Internet User,

Are you one of those people who love beautiful things around you?

Why not invest in an original oil painting?

For more information reply to:

ART@mfcnet.net

Sincerely Yours,

Phil & Jackie Clark

If you would like to be removed from our list, please send an
Email to remove@mfcnet.net with "remove" in the subject line.




From kaffe@kaffe.org  Sat Oct 25 23:24:13 1997
From: kaffe@kaffe.org (Matthias Hopf (@Home))
Date: Sat, 25 Oct 1997 22:24:13 MET1
Subject: NetBSD-m68k-0.9.1 compile problem
In-Reply-To: <199710221957.UAA18996@home.virtual-pc.com>
Message-ID: <yam7237.651.123577448@faui01.informatik.uni-erlangen.de>

Hi Nikolai!

> >I'm running NetBSD-m68k-hp300 1.2E
> >
> this question was answered in this list already, but anyway, i couldnt run
> kaffe 9.1 on my Mac, due to problem with symbols of some libraries, this 
> was
> also discussed in this list, but when i compiled it, and ran, it told me 
> about some
> symbols problem. I didnt insist, but if you make it work, let me know i 
> will fight again.

Sorry, I don't have a NextStep computer and thus I am not able to help you.
I only wrote the original m68k-assembler macros (for AmigaOS) which were
used for the NextStep port. Best ask at the mailing list kaffe@kaffe.org
(majordomo) for help.

Don't know why I stand in NOTES/HELPERS as NeXT porter :-]

Sorry

Matthias

-- 
    //             |   Matthias Hopf - "Hoeppel"   |    _      __
\\ //    Amiga     |  student of computer science  |  _|cience |-iction
 \X/ by conviction |      in Erlangen/Germany      | by belief in Future

 EMail:  mshopf@informatik.uni-erlangen.de
 Aminet: ftpamiga@epix.rrze.uni-erlangen.de
 WWW:    http://wwwcip.informatik.uni-erlangen.de/user/mshopf/


From kaffe@kaffe.org  Sun Oct 26 01:53:06 1997
From: kaffe@kaffe.org (Duane C. Cato)
Date: Sat, 25 Oct 1997 20:53:06 -0400
Subject: A Beautiful Painting!
Message-ID: <3.0.32.19971025205240.012980d4@mail.geocities.com>

At 02:08 AM 10/25/97 +0100, you wrote:
>Dear Internet User,
>
>Are you one of those people who love beautiful things around you?
>
>Why not invest in an original oil painting?
>
>For more information reply to:
>
>ART@mfcnet.net
>
>Sincerely Yours,
>
>Phil & Jackie Clark
>
>If you would like to be removed from our list, please send an
>Email to remove@mfcnet.net with "remove" in the subject line.

Look - PHIL!
Don't Spam, it is annoying, and simply uses up our listserver space. If we
want your stuff, we can and WILL find you.
Remove this listserver address from your list!

From kaffe@kaffe.org  Sun Oct 26 14:26:23 1997
From: kaffe@kaffe.org (Matthias Hopf (@Home))
Date: Sun, 26 Oct 1997 14:26:23 MET1
Subject: NetBSD-m68k-0.9.1 compile problem
In-Reply-To: <yam7237.651.123577448@faui01.informatik.uni-erlangen.de>
Message-ID: <yam7238.1178.124108720@faui01.informatik.uni-erlangen.de>

Hihi!

Matthias wrote:

> Sorry, I don't have a NextStep computer and thus I am not able to help you.
> I only wrote the original m68k-assembler macros (for AmigaOS) which were
> used for the NextStep port. Best ask at the mailing list kaffe@kaffe.org
> (majordomo) for help.

Grrr, should take a look at the header next time I'm replying to a
message... :-]

> Don't know why I stand in NOTES/HELPERS as NeXT porter :-]

This is still correct, though. I thought I should send a message to Tim
after my final exams (tuesday), but then, I can do it immedeately...

Tim, could you please remove me as NeXT porter in NOTES/HELPERS? I'm
repeatedly getting questions about the NeXT port, and I can't help those
folks. I don't know, who's done the NeXT port, either...

I'm still responsible for the Amiga port, though.

Cheers

Matthias

-- 
    //             |   Matthias Hopf - "Hoeppel"   |    _      __
\\ //    Amiga     |  student of computer science  |  _|cience |-iction
 \X/ by conviction |      in Erlangen/Germany      | by belief in Future

 EMail:  mshopf@informatik.uni-erlangen.de
 Aminet: ftpamiga@epix.rrze.uni-erlangen.de
 WWW:    http://wwwcip.informatik.uni-erlangen.de/user/mshopf/


From kaffe@kaffe.org  Sun Oct 26 17:55:15 1997
From: kaffe@kaffe.org (Martin Lanzen)
Date: Sun, 26 Oct 1997 18:55:15 +0100 (MET)
Subject: remove
In-Reply-To: <199710250108.CAA25266@cripplecock.sarc.city.ac.uk>
Message-ID: <199710261755.SAA14533@puh>

On 25 Okt , wwp@mfcnet.net wrote:
> Dear Internet User,
> 
> Are you one of those people who love beautiful things around you?
> 
> Why not invest in an original oil painting?
> 
> For more information reply to:
> 
> ART@mfcnet.net
> 
> Sincerely Yours,
> 
> Phil & Jackie Clark
> 
> If you would like to be removed from our list, please send an
> Email to remove@mfcnet.net with "remove" in the subject line.
> 
> 
> 
> 

-- 
                                                           
/Martin                                          \ \ | / /
                                                  ( @ @ )
--------------------------------------------o00o----()----o00o-----
| Martin Lanzen                                                      |
| Phone: 0303-19384,031-7036178,0707-414757                          |
| OBS --> New adress: Åsgatan 16, 442 30 Kungälv  <-- OBS            |
| Email: e4lanzen@etek.chalmers.se              .oooO                |
| http://www.etek.chalmers.se/~e4lanzen/         (   )   Oooo.       |
--------------------------------------------------\ (----(   )--------
                                                   \_)    ) /
                                                         (_/


From kaffe@kaffe.org  Sun Oct 26 20:04:16 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Mon, 27 Oct 1997 07:04:16 +1100
Subject: OtS Music Mag & Classifieds
Message-ID: <199710262004.HAA00917@Hi-Fi.infinet.net.au.>

Dear Member,

Edition 871 of the music Mag is on line.

Classifieds are now working from the site with over 50

classifications at http://www.onthestreet.com

There is no charge for using our classified service.

You can advertise anything including Gigs.

If you want to be deleted from our member service send an email 
with "remove" in the subject.

Thanks,

On The Street Magazine
Phone 02 9211 1122
Australia
From kaffe@kaffe.org  Mon Oct 27 06:59:23 1997
From: kaffe@kaffe.org (Danilov Nikita)
Date: Mon, 27 Oct 1997 09:59:23 +0300
Subject: remove
References: <199710262004.HAA00917@Hi-Fi.infinet.net.au.>
Message-ID: <34543BCB.38DA7CC8@server.ru>

This is a multi-part message in MIME format.
--------------CF7FE7283E109198A5AD69FD
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

ots@infinet.net.au wrote:
> 
> Dear Member,
> 
> Edition 871 of the music Mag is on line.
> 
> Classifieds are now working from the site with over 50
> 
> classifications at http://www.onthestreet.com
> 
> There is no charge for using our classified service.
> 
> You can advertise anything including Gigs.
> 
> If you want to be deleted from our member service send an email
> with "remove" in the subject.
> 
> Thanks,
> 
> On The Street Magazine
> Phone 02 9211 1122
> Australia

-- 
Mail me: god@server.ru                  Phone  me:  +7(095)4348205
Web  me: http://finn.server.ru/~god     Fax    me:  +7(095)4349267
         http://www.server.ru/~god
Of the first was he to bare arms and a name: 
Wassaily Booslaeugh of Riesengeborg. // J.Joyce, Finnegans Wake
--------------CF7FE7283E109198A5AD69FD
Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Nikita Danilov
Content-Disposition: attachment; filename="vcard.vcf"

begin:          vcard
fn:             Nikita Danilov
n:              Danilov;Nikita
org:            Server Ltd.
adr:            Server Ltd.,;;78 Vernardsky pr.;Moscow;Moscow region;117454;Russia
email;internet: god@server.ru
title:          Expert
tel;work:       +7(095)4348205
tel;fax:        +7(095)4348205
x-mozilla-cpt:  ;0
x-mozilla-html: TRUE
version:        2.1
end:            vcard


--------------CF7FE7283E109198A5AD69FD--

From kaffe@kaffe.org  Mon Oct 27 09:23:08 1997
From: kaffe@kaffe.org (Danilov Nikita)
Date: Mon, 27 Oct 1997 12:23:08 +0300
Subject: remove
References: <199710262004.HAA00917@Hi-Fi.infinet.net.au.> <34543BCB.38DA7CC8@server.ru>
Message-ID: <34545D7C.187335A4@server.ru>

This is a multi-part message in MIME format.
--------------AA55EB946375B9432B39F822
Content-Type: text/plain; charset=koi8-r
Content-Transfer-Encoding: 7bit

Sorry, next time I shall see to the headers thoroughly

Danilov Nikita wrote:
> 
> ots@infinet.net.au wrote:
> >
> > Dear Member,
...

Nikita.
-- 
Mail me: god@server.ru                  Phone  me:  +7(095)4348205
Web  me: http://finn.server.ru/~god     Fax    me:  +7(095)4349267
         http://www.server.ru/~god
Of the first was he to bare arms and a name: 
Wassaily Booslaeugh of Riesengeborg. // J.Joyce, Finnegans Wake
--------------AA55EB946375B9432B39F822
Content-Type: text/x-vcard; charset=koi8-r; name="vcard.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Nikita Danilov
Content-Disposition: attachment; filename="vcard.vcf"

begin:          vcard
fn:             Nikita Danilov
n:              Danilov;Nikita
org:            Server Ltd.
adr:            Server Ltd.,;;78 Vernardsky pr.;Moscow;Moscow region;117454;Russia
email;internet: god@server.ru
title:          Expert
tel;work:       +7(095)4348205
tel;fax:        +7(095)4348205
x-mozilla-cpt:  ;0
x-mozilla-html: TRUE
version:        2.1
end:            vcard


--------------AA55EB946375B9432B39F822--

From kaffe@kaffe.org  Mon Oct 27 09:02:05 1997
From: kaffe@kaffe.org (Umaire benoit)
Date: Mon, 27 Oct 1997 10:02:05 +0100
Subject: remove
Message-ID: <199710270902.KAA00639@sun46.butt1>

remove

From kaffe@kaffe.org  Mon Oct 27 11:19:28 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Mon, 27 Oct 1997 22:19:28 +1100 (EST)
Subject: java.util.Date not compatible b/w kaffe-0.8.4 and 0.9.1
Message-ID: <199710271119.WAA17776@zen.quick.com.au>

It looks very much like there is something very different b/w kaffe
versions wrt Date native methods.

I have an Strftime() class which provides decent formatting for times
etc.  It currently needs to be compiled with jdk1.0.2 (kaffe-0.8.4)
and when run with that VM produces correct results:

sjg:771$ JDK=1.0.2 java au.com.quick.util.Strftime %C "%Y%m%d %T"       
default: Mon Oct 27 22:16:16  GMT-1100 1997
%C: Mon Oct 27 22:16:16  GMT-1100 1997
%Y%m%d %T: 19971027 22:16:16
sjg:772$ date
Mon Oct 27 22:16:17 EST 1997

If I run it with kaffe-0.9.1 though I get:

sjg:773$ java au.com.quick.util.Strftime %C "%Y%m%d %T"           
default: Mon Oct 27 11:17:06  GMT 1997
%C: Mon Oct 27 11:17:06  GMT 1997
%Y%m%d %T: 19971027 11:17:06
sjg:774$ date
Mon Oct 27 22:17:07 EST 1997

If JDK is not set, my java_wrapper uses the greatest installed jdk
(1.1.4 here)

I would guess that the TimeZone() stuff does not work in 0.9.1
[sorry havn't solved the core dumps in 0.9.2 so can't test it].

--sjg

From kaffe@kaffe.org  Mon Oct 27 14:25:30 1997
From: kaffe@kaffe.org (Tim Wilkinson)
Date: Mon, 27 Oct 1997 14:25:30 +0000
Subject: java.util.Date not compatible b/w kaffe-0.8.4 and 0.9.1
References: <199710271119.WAA17776@zen.quick.com.au>
Message-ID: <3454A45A.167EB0E7@tjwassoc.co.uk>

Hi,

In JDK 1.0.2 java.util.Date used a number of native methods (Kaffe
0.8.x).  But in JDK 1.1.x java.util.Date is completely written in Java
and has not Kaffe bits at all (kaffe 0.9.x).

Tim
--
  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
  London, UK.                           Email:   tim@tjwassoc.co.uk

From kaffe@kaffe.org  Mon Oct 27 19:18:20 1997
From: kaffe@kaffe.org (Mark Hamzy)
Date: Mon, 27 Oct 1997 14:18:20 -0500 (EST)
Subject: remove
In-Reply-To: <199710250108.CAA25266@cripplecock.sarc.city.ac.uk> from "wwp@mfcnet.net" at Oct 25, 97 02:08:49 am
Message-ID: <199710271918.OAA12838@grumpy.magg.net>

remove

> 
> Dear Internet User,
> 
> Are you one of those people who love beautiful things around you?
> 
> Why not invest in an original oil painting?
> 
> For more information reply to:
> 
> ART@mfcnet.net
> 
> Sincerely Yours,
> 
> Phil & Jackie Clark
> 
> If you would like to be removed from our list, please send an
> Email to remove@mfcnet.net with "remove" in the subject line.
> 
> 
> 
> 


-- 
Anyone failing to display the mandatory black velvet Methodist style painting
   of President Quayle on their living room wall... WILL BE SHOT.
Defenestration n.: [ModL defenestratio < L de + fenestra, window]
   a throwing out through a window

From kaffe@kaffe.org  Mon Oct 27 18:25:58 1997
From: kaffe@kaffe.org (Tim Wilkinson)
Date: Mon, 27 Oct 1997 18:25:58 +0000
Subject: [TEST] New majordomo
Message-ID: <3454DCB6.41C67EA6@tjwassoc.co.uk>

This is a test message since we've just moved the majorodomo to w3.org.

Tim
--
  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
  London, UK.                           Email:   tim@tjwassoc.co.uk

From kaffe@kaffe.org  Mon Oct 27 21:41:24 1997
From: kaffe@kaffe.org (Mary Valente)
Date: Mon, 27 Oct 1997 13:41:24 -0800
Subject: remove
References: <199710270902.KAA00639@sun46.butt1>
Message-ID: <34550A84.D20097AE@nbn.com>

(U) maire benoit wrote:
> 
> remove
O got your email. I'm just trying to get off anything regarding this
kaffe stuff. I got on by mistake. I have no idea what it is. THANKS>

From kaffe@kaffe.org  Mon Oct 27 22:00:33 1997
From: kaffe@kaffe.org (David Wozmak)
Date: Mon, 27 Oct 1997 18:00:33 -0400
Subject: remove
Message-ID: <v01530501b07abdbfd397@[192.68.0.2]>

Hey, folks!

I'd like to poke my pointy little head in here and remind everyone that,
since this is a listserv, posting a "remove" message to this list to
respond to the spam that came in here, is not going to accomplish anything.
Nor will sending a "remove" message to the spammer with your email
address.  The spammer has 'kaffe@kaffe.org' listed as the recipient, and
doesn't know a gosh-derned thing about all of you individually.

and if any of the seven or eight of you  that DID reply to the list with
remove messages, wants to be removed from the listserv, you should follow
these directions:

If you ever want to remove yourself from this mailing list,
send the following command in email to
<kaffe-request@soi.city.ac.uk>:

    unsubscribe

Or you can send mail to <Majordomo@soi.city.ac.uk> with the following
command in the body of your email message:

    unsubscribe kaffe your_name@your_provider.



>remove
>
>>
>> Dear Internet User,
>>
>> Are you one of those people who love beautiful things around you?
>>
>> Why not invest in an original oil painting?
>>
>> For more information reply to:
>>


From kaffe@kaffe.org  Mon Oct 27 22:43:54 1997
From: kaffe@kaffe.org (Bruce K. Partlow)
Date: Mon, 27 Oct 1997 14:43:54 -0800
Subject: remove
Message-ID: <01BCE2E6.C019C100@brucep.datapac.com>

remove

From kaffe@kaffe.org  Sun Oct 26 15:14:56 1997
From: kaffe@kaffe.org (Mikael Ståldal)
Date: Sun, 26 Oct 1997 16:14:56 +0100
Subject: Kaffe 0.9.2: error in kaffe/kaffe/Makefile
Message-ID: <w51U0EpfhGyX092yn@ingen.reklam.nada.kth.se>

When the kaffe executable is build, the LDFLAGS variable isn't used. Why
not? It's used when building kaffeh.

And in the same Makefile, make clean tries to remove $(LIB) which isn't
defined, not good.

Here is a patch:

*** E:\unix\kaffe-0.9.2\kaffe\kaffe\Makefile.in	Thu Jun 12 06:11:36 1997
--- Makefile.in	Sun Oct 26 20:03:10 1997
***************
*** 20,25 ****
--- 20,26 ----
  INCLUDES=	-I. -I$(srcdir)/../kaffevm -I../../config -I$(srcdir)/../../config -I../../include -I$(srcdir)/../../include
  CFLAGS=		@CFLAGS@
  ALL_CFLAGS=	$(CFLAGS) @KAFFEFLAGS@ $(INCLUDES)
+ LDFLAGS=	@LDFLAGS@
  LIBS=		@NOSHAREDLIBS@
  OBJEXT=		@OBJEXT@
  EXEEXT=		@EXEEXT@
***************
*** 29,41 ****
  all:		kaffe$(EXEEXT)
  
  kaffe$(EXEEXT):	$(OBJECT)
! 		$(CC) $(ALL_CFLAGS) -o kaffe$(EXEEXT) $(OBJECT) $(LIBS)
  
  .c$(OBJEXT):
  		$(CC) $(ALL_CFLAGS) -c $<
  
  clean:
! 		rm -f $(LIB) kaffe$(EXEEXT) kaffe.core $(OBJECT)
  
  distclean:	clean
  		rm -f Makefile
--- 30,42 ----
  all:		kaffe$(EXEEXT)
  
  kaffe$(EXEEXT):	$(OBJECT)
! 		$(CC) $(LDFLAGS) -o kaffe$(EXEEXT) $(OBJECT) $(LIBS)
  
  .c$(OBJEXT):
  		$(CC) $(ALL_CFLAGS) -c $<
  
  clean:
! 		rm -f kaffe$(EXEEXT) kaffe.core $(OBJECT)
  
  distclean:	clean
  		rm -f Makefile

-- 
/****************************************************************\
* You have just read a message from Mikael Ståldal.              *
*                                                                *
* Remove "ingen.reklam." from the address before mail replying.  *
\****************************************************************/

From kaffe@kaffe.org  Sun Oct 26 18:52:15 1997
From: kaffe@kaffe.org (Mikael Ståldal)
Date: Sun, 26 Oct 1997 19:52:15 +0100
Subject: Kaffe 0.9.2 on OS/2
Message-ID: <fF5U0EpfhK4W092yn@ingen.reklam.nada.kth.se>

Here are some patches for OS/2. To get the config/i386/os2 direcrory, just
copy config/i386/win32 and rename the cygnus subdirectory to emx.

With this patches and EMX 0.9c its possible to build Kaffe on OS/2. It
doesn't work though, it crashes as soon as I try to use it. I tried both
interpreter and JIT.

Only in kaffe/config/i386: os2
diff -cbr kaffe-0.9.2/config.guess kaffe/config.guess
*** kaffe-0.9.2/config.guess	Thu Jun 26 08:20:10 1997
--- kaffe/config.guess	Sun Oct 26 18:47:58 1997
***************
*** 442,447 ****
--- 442,450 ----
      i*86:CYGWIN32/*:*:*)
  	echo i386-unknown-win32/cygnus
  	exit 0 ;;
+     i*86:OS/2:*:*)
+ 	echo i386-pc-os2/emx
+ 	exit 0 ;;
  
  esac
  
diff -cbr kaffe-0.9.2/config.sub kaffe/config.sub
*** kaffe-0.9.2/config.sub	Thu Jun 26 08:20:12 1997
--- kaffe/config.sub	Sun Oct 26 18:45:38 1997
***************
*** 697,702 ****
--- 697,704 ----
  	-xenix)
  		os=-xenix
  		;;
+         -os2*)
+ 		;;
  	-none)
  		;;
  	*)
diff -cbr kaffe-0.9.2/configure.in kaffe/configure.in
*** kaffe-0.9.2/configure.in	Sat Oct  4 03:26:52 1997
--- kaffe/configure.in	Tue Oct 21 22:42:48 1997
***************
*** 141,146 ****
--- 141,155 ----
  	LN_S="cp"
  	;;
  
+ i[3456]86-os2/emx)
+ 	host_cpu=i386
+ 	host_os=os2/emx
+ 	EXEEXT=.exe
+ 	SCREXT=.cmd
+ 	dynamic_libraries=no
+ 	LN_S="cp"
+ 	;;
+ 
  i[3456]86-dgux*)
  	host_cpu=i386
  	host_os=dgux
***************
*** 457,462 ****
--- 466,473 ----
  case $host_os in
  win32*)
  	SYSTEM=win32 ;;
+ os2*)
+ 	SYSTEM=os2 ;;
  *)
  	SYSTEM=unix ;;
  esac
***************
*** 1060,1065 ****
--- 1071,1085 ----
  	fi
  	echo set KAFFEHOME=$datadir/kaffe >> ENVIRONMENT.BAT
  	echo set PATH=%PATH%\;$bindir >> ENVIRONMENT.BAT
+ elif test "$SYSTEM" = "os2" ; then
+ 	echo @echo off > ENVIRONMENT.CMD
+ 	if test "$awt_toolkit" != "none" ; then
+ 		echo set CLASSPATH=.\;$datadir/kaffe/classes.zip\;$datadir/kaffe/$awt_toolkit >> ENVIRONMENT.CMD
+ 	else
+ 		echo set CLASSPATH=.\;$datadir/kaffe/classes.zip >> ENVIRONMENT.CMD
+ 	fi
+ 	echo set KAFFEHOME=$datadir/kaffe >> ENVIRONMENT.CMD
+ 	echo set PATH=%PATH%\;$bindir >> ENVIRONMENT.CMD
  fi
  
  dnl -------------------------------------------------------------------------
diff -cbr kaffe-0.9.2/include/system.h kaffe/include/system.h
*** kaffe-0.9.2/include/system.h	Mon May  5 07:25:32 1997
--- kaffe/include/system.h	Sun Oct 26 15:05:22 1997
***************
*** 22,28 ****
  #define	file_seperator		"/"
  #define	path_seperator		":"
  #define	line_seperator		"\n"
! #elif defined(__WIN32__) || defined(__OS2__)
  #define	file_seperator		"\\"
  #define	path_seperator		";"
  #define	line_seperator		"\r\n"
--- 22,28 ----
  #define	file_seperator		"/"
  #define	path_seperator		":"
  #define	line_seperator		"\n"
! #elif defined(__WIN32__) || defined(__OS2__) || defined(__EMX__)
  #define	file_seperator		"\\"
  #define	path_seperator		";"
  #define	line_seperator		"\r\n"
diff -cbr kaffe-0.9.2/kaffe/kaffevm/paths.h kaffe/kaffe/kaffevm/paths.h
*** kaffe-0.9.2/kaffe/kaffevm/paths.h	Wed Apr 16 01:10:56 1997
--- kaffe/kaffe/kaffevm/paths.h	Sun Oct 26 15:07:30 1997
***************
*** 18,24 ****
  #define	MAXPATHLEN	1024
  #endif
  
! #if defined(__WIN32__) || defined(__OS2__)
  #define	PATHSEP		';'
  #define	DIRSEP		"\\"
  #elif defined(__amigaos__)
--- 18,24 ----
  #define	MAXPATHLEN	1024
  #endif
  
! #if defined(__WIN32__) || defined(__OS2__) || defined(__EMX__)
  #define	PATHSEP		';'
  #define	DIRSEP		"\\"
  #elif defined(__amigaos__)

-- 
/****************************************************************\
* You have just read a message from Mikael Ståldal.              *
*                                                                *
* Remove "ingen.reklam." from the address before mail replying.  *
\****************************************************************/

From kaffe@kaffe.org  Mon Oct 27 21:45:54 1997
From: kaffe@kaffe.org (Mikael Ståldal)
Date: Mon, 27 Oct 1997 22:45:54 +0100
Subject: Problem with Makefiles: SHELL missing
Message-ID: <SuQV0Epfha0Y092yn@nada.kth.se>

The Makefiles make use of UNIX shell features, but they doesn't contain any
SHELL caluse. They should have a

SHELL = /bin/sh

line.

-- 
/****************************************************************\
* You have just read a message from Mikael Ståldal.              *
*                                                                *
* Remove "ingen.reklam." from the address before mail replying.  *
\****************************************************************/

From kaffe@kaffe.org  Tue Oct 28 05:37:30 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Tue, 28 Oct 1997 16:37:30 +1100 (EST)
Subject: java.util.Date not compatible b/w kaffe-0.8.4 and 0.9.1
References: <199710271119.WAA17776@zen.quick.com.au> <3454A45A.167EB0E7@tjwassoc.co.uk>
Message-ID: <199710280537.QAA15984@gate.quick.com.au>

>In JDK 1.0.2 java.util.Date used a number of native methods (Kaffe
>0.8.x).  But in JDK 1.1.x java.util.Date is completely written in Java
>and has not Kaffe bits at all (kaffe 0.9.x).

Thanks. Turns out that for jdk1.1.x you need to have user.timezone
set.  Naffed if I can find any doco which says what file java looks
in for that sort of thing... by default that is.

Oh if only they had kept getenv :-)

--sjg
-- 
Simon J. Gerraty        <sjg@quick.com.au>

#include <disclaimer>   /* imagine something _very_ witty here */

From kaffe@kaffe.org  Tue Oct 28 07:56:34 1997
From: kaffe@kaffe.org (Holger Veit)
Date: Tue, 28 Oct 1997 08:56:34 +0100 (MET)
Subject: Kaffe 0.9.2 on OS/2
In-Reply-To: <fF5U0EpfhK4W092yn@ingen.reklam.nada.kth.se> from "Mikael St\eldal" at Oct 26, 97 07:52:15 pm
Message-ID: <199710280756.AA00548@simi.gmd.de>

> 
> Here are some patches for OS/2. To get the config/i386/os2 direcrory, jus=
> t
> copy config/i386/win32 and rename the cygnus subdirectory to emx.
> 
> With this patches and EMX 0.9c its possible to build Kaffe on OS/2. It
> doesn't work though, it crashes as soon as I try to use it. I tried both
> interpreter and JIT.

Congratulations: you have likely found the same problem than I quite some
time ago, however I haven't bothered to try out the interpreter. So my
initial assumption of an obscure problem in the JIT or JIT-generated code
seems to be wrong. This makes room for other speculation: kaffe emulates
threading, rather than relying on native threads or pthread packages. I
had annotated the class loader and found the system crashing in different
classes which might look like a timing problem. I haven't seen a native
threads application in the tree yet (however this was 0.9.1, maybe this has
changed in 0.9.2), so a good idea might be to replace the internal threading
stuff with native threads. You can find a snapshot of my current 0.91 tree
at ftp://set.gmd.de/pub/misc/XFree86OS2/test/kaffe.zip for reference
(508K, directory readable; note it is not worthwhile to download it unless one 
is really interested in porting - it doesn't work).

-- 
         Dr.-Ing. Holger Veit             | INTERNET: Holger.Veit"at"gmd.de
|  |   / GMD - German National Research   | Phone: (+49) 2241 14 2448
|__|  /  Center for Information Technology| Fax:   (+49) 2241 14 2242
|  | /   Schloss Birlinghoven             |  Get XFree86/OS2 Bugfix Version
|  |/    D-53754 Sankt Augustin, Germany  |    V3.3.1 from ftp.xfree86.org
         WWW: http://borneo.gmd.de/~veit/ | /pub/XFree86/3.3.1/binaries/OS2

From kaffe@kaffe.org  Tue Oct 28 15:16:07 1997
From: kaffe@kaffe.org (Robert Bates)
Date: Tue, 28 Oct 1997 09:16:07 -0600
Subject: QNX 4.x port of 0.9.2...?
Message-ID: <3.0.3.32.19971028091607.0093b7d0@zeh.com>

Has anyone had any more luck getting kaffe 0.9.2 to port to QNX?  I'm about
to go out and sink the $$$ into the QNX devel system, and would like to get
a Java VM up and running ASAP on that platform.  Seeing as how QNX has a
few more months to go on their *own* JVM... ;)

I read the article about getting 0.9.1 to build on QNX, so I'm good at
least up to that point...

Also, are there any mirrors of the gcc port to QNX referenced in the
posting?  I'm having a hard time getting a complete d/l from the France
site...

TIA,
Rob

//////////////////////////////////////////////////////////////////////////
// Robert Bates
// Infosys Development Group
// Houston, Texas  USA
//
// rbates@iname.com
// http://home.att.net/~rpbates
//////////////////////////////////////////////////////////////////////////
#ifdef _LITIGATORS_PRESENT_
#include <std_disclaimer.h>
#endif

From kaffe@kaffe.org  Tue Oct 28 17:29:49 1997
From: kaffe@kaffe.org (Robert Bates)
Date: Tue, 28 Oct 1997 11:29:49 -0600
Subject: Kaffe 0.9.2 on QNX 4.x...?
Message-ID: <3.0.3.32.19971028112949.009824d0@zeh.com>

Has anyone had any more luck getting kaffe 0.9.2 to port to QNX?  I'm about
to go out and sink the $$$ into the QNX devel system, and would like to get
a Java VM up and running ASAP on that platform.  Seeing as how QNX has a
few more months to go on their *own* JVM... ;)

I read the article about getting 0.9.1 to build on QNX, so I'm good at
least up to that point...

Also, are there any mirrors of the gcc port to QNX referenced in the
posting?  I'm having a hard time getting a complete d/l from the France
site...

TIA,
Rob

//////////////////////////////////////////////////////////////////////////
// Robert Bates
// Infosys Development Group
// Houston, Texas  USA
//
// rbates@iname.com
// http://home.att.net/~rpbates
//////////////////////////////////////////////////////////////////////////
#ifdef _LITIGATORS_PRESENT_
#include <std_disclaimer.h>
#endif

From kaffe@kaffe.org  Tue Oct 28 18:11:09 1997
From: kaffe@kaffe.org (Yuang Shuang-Luang)
Date: Wed, 29 Oct 1997 02:11:09 +0800 (CST)
Subject: Failed to locate native function: getFontPeer
Message-ID: <199710281811.CAA10185@Edward.net>


>Submitter-Id:  kaffe-user
>Originator: 	Yuang Shuang-Luang
>Organization:	
>Priority:	medium
>Category: 	
>Class:		sw-bug 
>Release:	
>Environment:

	System: FreeBSD Edward 3.0-CURRENT FreeBSD 3.0-CURRENT #0: Mon Oct 27 18:55:20 CST 1997     edward@Edward:/usr/src/sys/compile/Ed  i386
	Architecture: i386
	Machine: pentium
>Description:

	Thw X-win version I used is 3.3.1, when I try to run appletviewer,
	it doesn't work and show the following error message:

	Failed to locate native function:
	java/awt/Toolkit.getFontPeer(Ljava/lang/String;I)Ljava/awt/peer/FontPeer
	;
	java.lang.UnsatisfiedLinkError
	at java/awt/Font.<init>(119)
	at sun/applet/AppletCopyright.<init>(40)
	at sun/applet/AppletViewer.mainInit(984)
	at sun/applet/AppletViewer.main(993)

	How can I make appletviewer work? Should I set some environment 
	variable to fix it? Thanx...

From kaffe@kaffe.org  Tue Oct 28 20:28:16 1997
From: kaffe@kaffe.org (Mary Valente)
Date: Tue, 28 Oct 1997 12:28:16 -0800
Subject: please remove!
References: <199710271918.OAA12838@grumpy.magg.net>
Message-ID: <34564AE0.9FAF7F02@nbn.com>

Mark Hamzy wrote:
> 
> remove
> 
> >
> > Dear Internet User,
> >
> > Are you one of those people who love beautiful things around you?
> >
> > Why not invest in an original oil painting?
> >NO>>>>>PLEASE REMOVE!
> > For more information reply to:
> >
> > ART@mfcnet.net
> >
> > Sincerely Yours,
> >
> > Phil & Jackie Clark
> >
> > If you would like to be removed from our list, please send an
> > Email to remove@mfcnet.net with "remove" in the subject line.
> >
> >
> >
> >
> 
> --
> Anyone failing to display the mandatory black velvet Methodist style painting
>    of President Quayle on their living room wall... WILL BE SHOT.
> Defenestration n.: [ModL defenestratio < L de + fenestra, window]
>    a throwing out through a window

From kaffe@kaffe.org  Wed Oct 29 16:09:16 1997
From: kaffe@kaffe.org (Daniel Veillard)
Date: Wed, 29 Oct 1997 11:09:16 -0500
Subject: Kaffe bug on PC
Message-ID: <199710291609.LAA23851@rufus.w3.org>


>I was able to build kaffe under VC++ 5.0. It does not
>work yet but looks promissing. I have the following
>problem:
>
>In initialization time while executing initialization
>of system class the vm gets into nested calls to
>itself situation. In file kaffe.def:
>
>define_insn(INVOKESTATIC)
>{
>......
>
>#if defined(TRANSLATOR) && defined(HAVE_TRAMPOLINE))
>  
>call_indirect_const(&METHOD_NATIVECODE(method_method());
>#else
>   call(tmp);
>#endif
>
>........
>}
>
>In my configuration the call(tmp) is what compiled
>and, if you look at this macro it is, basicly, calling
>virtualMachine() which again gets back to INVOKESTATIC
>case and like that - forever until the program runs
>out of stack.
>
>Any clue????

   Not really, did you try to compile the interpreter or the JIT.
BTW I'm very interested in a version of Kaffe in the VC++ environment.
Did you had to patch a lot of things ?

Daniel


From kaffe@kaffe.org  Wed Oct 29 17:30:13 1997
From: kaffe@kaffe.org (Michael Gesundheit)
Date: Wed, 29 Oct 1997 09:30:13 -0800 (PST)
Subject: Kaffe bug on PC
Message-ID: <19971029173013.23424.rocketmail@web1.rocketmail.com>


I build kaffe for interpreter. Not too many patches.
It is not running yet. My current problem has to to
with the slots union.

Luckyly I have a Solaris Ultra next to my PC. I run
gdb on Solaris and VC++ on the PC.

The problem has to do with HelloWorldApp.class. The 4th
byte code is INVOKEVIRTUAL. On the PC one of the macros
produce a null pointer which is basicly the 3rd stack
entry. It should, and is on Solaris, a pointer but
NULL on the PC. I didn't pinpoint the problem yet but
it has to do with the difference between the compilers
and the way they handle unions. The size of the slots
union is 8 bytes. I see the pointer in the upper 4
bytes in both the PC and Solaris. gcc takes the pointer
correctly. VC++ takes the lower 4 bytes which are NULL.
I'll address this problem today and broadcast the fix.
It smell like a fundamental issue for PC.

About kaffe in general - I built it initialy with DLL
but converted it to static libraries. I have this carzy
idea of running Kaffe on an embedded system which does
handle DLLs. In addition I will ROMize all the class
libraries to be loaded with the kaffe image as one
enchilada.

I'll probebly post progress along the way.

--Michael




---Daniel Veillard <veillard@rufus.w3.org> wrote:
>
> 
> >I was able to build kaffe under VC++ 5.0. It does
not
> >work yet but looks promissing. I have the following
> >problem:
> >
> >In initialization time while executing
initialization
> >of system class the vm gets into nested calls to
> >itself situation. In file kaffe.def:
> >
> >define_insn(INVOKESTATIC)
> >{
> >......
> >
> >#if defined(TRANSLATOR) && defined(HAVE_TRAMPOLINE))
> >  
>
>call_indirect_const(&METHOD_NATIVECODE(method_method());
> >#else
> >   call(tmp);
> >#endif
> >
> >........
> >}
> >
> >In my configuration the call(tmp) is what compiled
> >and, if you look at this macro it is, basicly,
calling
> >virtualMachine() which again gets back to
INVOKESTATIC
> >case and like that - forever until the program runs
> >out of stack.
> >
> >Any clue????
> 
>    Not really, did you try to compile the
interpreter or the JIT.
> BTW I'm very interested in a version of Kaffe in
the VC++ environment.
> Did you had to patch a lot of things ?
> 
> Daniel
> 
> 
> 
> 

_____________________________________________________________________
Sent by RocketMail. Get your free e-mail at http://www.rocketmail.com

From kaffe@kaffe.org  Wed Oct 29 18:05:42 1997
From: kaffe@kaffe.org (Aaron Coday)
Date: Wed, 29 Oct 1997 12:05:42 -0600
Subject: Kaffe bug on PC
References: <199710291609.LAA23851@rufus.w3.org>
Message-ID: <34577AF6.FF446DB9@cs.uiuc.edu>

Hi,

I have in the past worked on getting Kaffe running on win32/VC++, but
always run into a bunch of problems.  I think it would be very helpful
if when you get it working you could post some notes/diffs about
problems specific to win32/VC++.

thanks in advance.
-Aaron


Daniel Veillard wrote:
> 
> >I was able to build kaffe under VC++ 5.0. It does not
> >work yet but looks promissing. I have the following
> >problem:
> >
> >In initialization time while executing initialization
> >of system class the vm gets into nested calls to
> >itself situation. In file kaffe.def:
> >
> >define_insn(INVOKESTATIC)
> >{
> >......
> >
> >#if defined(TRANSLATOR) && defined(HAVE_TRAMPOLINE))
> >
> >call_indirect_const(&METHOD_NATIVECODE(method_method());
> >#else
> >   call(tmp);
> >#endif
> >
> >........
> >}
> >
> >In my configuration the call(tmp) is what compiled
> >and, if you look at this macro it is, basicly, calling
> >virtualMachine() which again gets back to INVOKESTATIC
> >case and like that - forever until the program runs
> >out of stack.
> >
> >Any clue????
> 
>    Not really, did you try to compile the interpreter or the JIT.
> BTW I'm very interested in a version of Kaffe in the VC++ environment.
> Did you had to patch a lot of things ?
> 
> Daniel

-- 
===============================================================
Aaron Coday				      a-coday@uiuc.edu
2234 DCL  MC 258  	http://www.students.uiuc.edu/~a-coday/
1304 W. Springfield
Urbana, IL  61801	"No matter where you go, there you are"
(W)217-244-7116
===============================================================

From kaffe@kaffe.org  Wed Oct 29 20:21:03 1997
From: kaffe@kaffe.org (Robert Bates)
Date: Wed, 29 Oct 1997 14:21:03 -0600
Subject: Possible licensing issue with JavaSoft?
Message-ID: <3.0.3.32.19971029142103.00947bf0@zeh.com>

I just got off the phone with QNX about porting Kaffe to QNX and using it
in a commercial product... They expressed concern over my using it on a
commercial platform, and even though Kaffe is a clean-room project, they
have heard rumblings from JavaSoft about quashing any VM running without a
JavaSoft license on a commercialized platform.

Has anyone got any info regarding the legalities behind these
fears/concerns?  Basically, I'm investigating using the Kaffe port on QNX
to build commercially-available servers, and would like to know if I'm
getting a bit optimistic (legally) about being able to sell the systems
with Kaffe loaded on them... QNX would love to know about it too, because
they are in the process of sinking serious $$$ into a JavaSoft license to
get a VM ported to run under QNX and Neutrino.

BTW, their's won't be available until mid '98, and they are actually urging
me to explore this avenue for an "immediate" Java solution... (I think they
have a vested interest in my product shipping ASAP... ;)

Thanks,
Rob
//////////////////////////////////////////////////////////////////////////
// Robert Bates
// Infosys Development Group
// Houston, Texas  USA
//
// rbates@iname.com
// http://home.att.net/~rpbates
//////////////////////////////////////////////////////////////////////////
#ifdef _LITIGATORS_PRESENT_
#include <std_disclaimer.h>
#endif

From kaffe@kaffe.org  Wed Oct 29 20:31:58 1997
From: kaffe@kaffe.org (John D. Gwinner)
Date: Wed, 29 Oct 1997 15:31:58 -0500
Subject: Kaffe bug on PC
Message-ID: <199710292019.UAA29778@home.virtual-pc.com>

Michael:
...
> it has to do with the difference between the compilers
> and the way they handle unions. The size of the slots
> union is 8 bytes. I see the pointer in the upper 4
> bytes in both the PC and Solaris. gcc takes the pointer
> correctly. VC++ takes the lower 4 bytes which are NULL.
> I'll address this problem today and broadcast the fix.
> It smell like a fundamental issue for PC.

Just to cover all bases, have you examined the packing (#pragma pack)?  I
don't think there is a guaranteed byte placement within a union, but I
could be wrong -- I'm just guessing.

Are you using INT64 for the 8 byte quantity to break out the 4 bytes? 
That might be one way to go.

		== John ==

From kaffe@kaffe.org  Wed Oct 29 20:50:02 1997
From: kaffe@kaffe.org (Rob Mayoff)
Date: Wed, 29 Oct 1997 14:50:02 -0600 (CST)
Subject: Possible licensing issue with JavaSoft?
In-Reply-To: <3.0.3.32.19971029142103.00947bf0@zeh.com>
Message-ID: <Pine.LNX.3.96.971029144658.20767D-100000@dqd.com>

+-- On Wed, 29 Oct 1997, Robert Bates wrote:
| I just got off the phone with QNX about porting Kaffe to QNX and using it
| in a commercial product... They expressed concern over my using it on a
| commercial platform, and even though Kaffe is a clean-room project, they
| have heard rumblings from JavaSoft about quashing any VM running without a
| JavaSoft license on a commercialized platform.

Read the copyright page of "The Java Language Specifications".  It
grants a license to implement the spec, but you have to do some things
like pass the test suites.

From kaffe@kaffe.org  Wed Oct 29 22:49:02 1997
From: kaffe@kaffe.org (Tony Kimball)
Date: Wed, 29 Oct 1997 16:49:02 -0600 (CST)
Subject: Possible licensing issue with JavaSoft?
References: <3.0.3.32.19971029142103.00947bf0@zeh.com>
Message-ID: <199710292249.QAA04668@compound.east.sun.com>

Quoth Robert Bates on Wed, 29 October:
: I just got off the phone with QNX about porting Kaffe to QNX and using it
: in a commercial product... They expressed concern over my using it on a
: commercial platform, and even though Kaffe is a clean-room project, they
: have heard rumblings from JavaSoft about quashing any VM running without a
: JavaSoft license on a commercialized platform.

Where have they heard this?

: Has anyone got any info regarding the legalities behind these
: fears/concerns? 

I'm not a lawyer but I've been sued in the past:-).  I am very much of
the opinion that if JavaSoft were to sue you for using Kaffe, you
would win a bunch of money in your countersuit.  Moreover, JavaSoft
people (not corporate officers) have told me that Java is an open
spec, and that such a suit is the farthest thing from their minds.
(Frankly, I think they'd lose a lot of crucial staff if they attacked
freeware.)  Finally, there are commerical products already in
existence using non-JavaSoft development kits and JVMs, from Japan,
and there is no sign of legal action from JavaSoft -- quite the
contrary, they seem to be actively cooperating on interface
specification issues for embedded systems.

On the otherhand, if you were to use the word "Java" you would be
inviting some sort of action, since they regard trademark dilution as
a serious threat.  "JV" or "J" are typical work-arounds, allowing you
to identify the functional characteristics of your product without
implying certifiable compliance.

: QNX would love to know about it too, because
: they are in the process of sinking serious $$$ into a JavaSoft license to
: get a VM ported to run under QNX and Neutrino.

That's the only way they can ever use the Java name, branding, or
logos, and have legal peace.  This has been JavaSoft's clearly
defended turf.

From kaffe@kaffe.org  Thu Oct 30 00:36:48 1997
From: kaffe@kaffe.org (Per Bothner)
Date: Wed, 29 Oct 1997 16:36:48 -0800
Subject: Possible licensing issue with JavaSoft?
In-Reply-To: Your message of "Wed, 29 Oct 1997 14:21:03 CST."
 <3.0.3.32.19971029142103.00947bf0@zeh.com>
Message-ID: <199710300035.QAA14937@cygnus.com>

> They expressed concern over my using it on a
> commercial platform, and even though Kaffe is a clean-room project, they
> have heard rumblings from JavaSoft about quashing any VM running without a
> JavaSoft license on a commercialized platform.

There are three separate issues:

a) Use of the trade-marked "Java" name.  In commercial use you would want
to be very careful when using "Java" without Sun's permission.  This is
an issue that came up when Sun proposed Java for ISO certification.
There is some Sun-approved work-around (something like "Java platform")
- look around java.sun.com.

2) Whether Sun would sue (or threaten to sue) over a commercial
clean-room re-implementation of Java, and if they did, whether they
would have a reasonable chance of prevailing.  I believe the answer
to these questions is no, and Cygnus is investing big bucks on that
belief.

3) Whether Kaffe is a clean-room implementation of Java.
Even though Kaffe is a clean-room re-implementation (or at least
close enough), the fact is that Kaffe is useless without Sun's
classes.zip.  And the license on classes.zip prevents its use
with Kaffe in a commercial context.  That is why a free replacement
for classes.zip is needed.  Cygnus will be re-implementing at
least the core classes.  (We have not looked at Kore or any
alternatives yet, so I cannot say to what extent we will
start with that.)  But we cannot re-implement all the features
in Sun's classes.zip in a timely manner (especially when Sun
keeps adding things like Java2D, Java3D, JDBC, etc etc), so
we hope "the Net" will help with the non-core classes.
(The core classes are too critical for us to wait for the net to
provide.)

	--Per Bothner
Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner

From kaffe@kaffe.org  Thu Oct 30 00:59:18 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Wed, 29 Oct 1997 16:59:18 -0800
Subject: Possible licensing issue with JavaSoft?
In-Reply-To: Your message of "Wed, 29 Oct 1997 16:36:48 PST."
 <199710300035.QAA14937@cygnus.com>
Message-ID: <199710300059.QAA20853@bach.marimba.com>


>a) Use of the trade-marked "Java" name.  In commercial use you would want
>to be very careful when using "Java" without Sun's permission.  This is
>an issue that came up when Sun proposed Java for ISO certification.
>There is some Sun-approved work-around (something like "Java platform")
>- look around java.sun.com.

What's the status of the "Unix" name?  Failing a Sun-approved 
work-around, I suppose that "J*va" would work, and be pretty useful, 
much like Un*x etc.

>2) Whether Sun would sue (or threaten to sue) over a commercial
>clean-room re-implementation of Java, and if they did, whether they
>would have a reasonable chance of prevailing.  I believe the answer
>to these questions is no, and Cygnus is investing big bucks on that
>belief.

I remember a tale of someone who put a Java logo on his homepage in 
the very early days of the language, to show his support.  But then 
Sun threatened to sue him...  I think that Sun's legal department has 
been quite stupidly aggressive in the past, and is probably still so.  
Whether or not they stand a chance of winning, they might try.  
Perhaps they were just bored back then, in which case we're all safe 
for now ;)

>3) Whether Kaffe is a clean-room implementation of Java.
>Even though Kaffe is a clean-room re-implementation (or at least
>close enough), the fact is that Kaffe is useless without Sun's
>classes.zip.

"The Net" plans to help out, I guess.  Whether they can keep up or not 
is another question, especially when Sun starts pulling a Microsoft 
and changing core classes gratuitously, as I expect they'll do when 
they feel comfortable about MS.  We will live in interesting times, 
indeed.

It would be kind of silly for any company _not_ to want a monopoly, 
wouldn't it?

-- 
benjamin@marimba.com            http://www.marimba.com/people/benjamin
PGP public key on my homepage, and finger -l bwpearre@cs.princeton.edu
1024/CC2761B9 97/2/11 E3 60 33 2F 1F 9D 17 60  DD 45 46 7E DF 1F 81 89 


From kaffe@kaffe.org  Thu Oct 30 02:06:52 1997
From: kaffe@kaffe.org (Michael Gesundheit)
Date: Wed, 29 Oct 1997 18:06:52 -0800 (PST)
Subject: Kaffe bug on PC
Message-ID: <19971030020652.24174.rocketmail@web2.rocketmail.com>






---"John D. Gwinner" <gwinner@northnet.org> wrote:
>
> Michael:
> ..
> > it has to do with the difference between the
compilers
> > and the way they handle unions. The size of the
slots
> > union is 8 bytes. I see the pointer in the upper 4
> > bytes in both the PC and Solaris. gcc takes the
pointer
> > correctly. VC++ takes the lower 4 bytes which are
NULL.
> > I'll address this problem today and broadcast the
fix.
> > It smell like a fundamental issue for PC.
> 
> Just to cover all bases, have you examined the
packing (#pragma pack)? 

No, under 0.9.2 the thread package does not require
assembly. setjmp/longjmp is used.
 I
> don't think there is a guaranteed byte placement
within a union, but I
> could be wrong -- I'm just guessing.
> 
> Are you using INT64 for the 8 byte quantity to
break out the 4 bytes? 

Yes.

> That might be one way to go.

The problem as I see it at this point is not a union
issue. I realized today that the problem comes from
the initial GETSTATIC opcode processing. The
move_ref_const macro moves the pointer
field->info.addr to tmp which is a slots pointer then
the macro load_ref move the contant, which is 0 (!!) to
the stack. Later on the bytecode INVOKEVIRTUAL takes
this stack entry as a pointer and crashes. At this 
point I think the bug is in initializing my 
HelloWorldApp class. This missing pointer should be 
set when the class is loaded and processed. I hope
that the Solaris I have next to the PC will help
tracing the class initialization.

--Michael
macro moves a field->info.addr pointer which points
to 0. The following move_ref_const

_____________________________________________________________________
Sent by RocketMail. Get your free e-mail at http://www.rocketmail.com

From kaffe@kaffe.org  Thu Oct 30 12:50:22 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Thu, 30 Oct 1997 13:50:22 +0100
Subject: Alpha linux hangs when using javac
Message-ID: <199710301250.NAA01488@marvinii.link.no>


>Submitter-Id:  kaffe-user
>Originator:    Petter Reinholdtsen
>Organization:  Telenor Link
>Confidential:  no
>Synopsis:      kaffe hangs when running javac
>Severity:      critical
>Priority:      high
>Category:      kaffe
>Class:         sw-bug
>Release:	0.81
>Environment:
        System: Linux xenu 2.0.30 #6 Tue May 13 19:31:06 CEST 1997 alpha unknown
        Architecture: alpha
        Machine: DEC
>Description:
	kaffe hangs when trying to compile something using javac.  It
	uses 90% CPU, but is never done.
>How-To-Repeat:
	After getting kaffe to compile on Linux Alpha (had to add
	-Wl,-rpath,/usr/local/lib to the linking of kaffe), I tried
	running javac to compile test/general/RandomTest.java.  I had
	to add CLASSPATH=/usr/local/share/kaffe/classes.zip and
	LD_LIBRARY_PATH=/usr/local/lib to environment to get it to run
	at all.  The compiler never finishes.
>Fix:
	Don't know.

From kaffe@kaffe.org  Thu Oct 30 14:01:19 1997
From: kaffe@kaffe.org (David Scott)
Date: Thu, 30 Oct 1997 09:01:19 -0500 (EST)
Subject: Possible licensing issue with JavaSoft?
In-Reply-To: <199710300059.QAA20853@bach.marimba.com>
Message-ID: <Pine.LNX.3.95.971030084932.11373D-100000@studmuffin.ganymede.com>


On Wed, 29 Oct 1997 benjamin@marimba.com wrote:
 
> I remember a tale of someone who put a Java logo on his homepage in 
> the very early days of the language, to show his support.  But then 
> Sun threatened to sue him...  I think that Sun's legal department has 
> been quite stupidly aggressive in the past, and is probably still so.  
> Whether or not they stand a chance of winning, they might try.  
> Perhaps they were just bored back then, in which case we're all safe 
> for now ;)

My take on this issue is this: Sun is being very careful in defining its
legal position with regard to Java(tm) because of Microsoft, whom Sun
doesn't want mucking around with Java the way they muck around with
everything else. The thing is, for them to defend their turf against the
Evil Empire, they have to enforce some silly-looking boilerplate even
among more friendly folk. 

For example, it appears that O'Reilly Associates has recently heard from
Sun's lawyers about using the Java name without the (tm) on the covers of
their excellent Java books. This is particularly bizarre when you consider
how Java-friendly ORA really is. 

For the time being, I think Sun should be given very wide berth on this
issue, until things sort themselves out a bit.  I find their lawsuit
against Microsoft very interesting.  One way to support them is simply to
play along. 

David Scott
Montreal

From kaffe@kaffe.org  Thu Oct 30 14:21:35 1997
From: kaffe@kaffe.org (Otto Hammersmith)
Date: Thu, 30 Oct 1997 09:21:35 -0500
Subject: Alpha linux hangs when using javac
In-Reply-To: <199710301250.NAA01488@marvinii.link.no>; from pere@td.org.uit.no on Thu, Oct 30, 1997 at 01:50:22PM +0100
References: <199710301250.NAA01488@marvinii.link.no>
Message-ID: <19971030092135.54643@redhat.com>

FWIW, something similar happens on an intel with glibc 2
running... I'd bet that it's glibc and not a 32-64 bit problem.  I'm
intending to take a look at it in the next couple of days, I'll
certainly forward a patch if I come up with anything.

On Thu, Oct 30, 1997 at 01:50:22PM +0100, pere@td.org.uit.no wrote:
> 
> >Submitter-Id:  kaffe-user
> >Originator:    Petter Reinholdtsen
> >Organization:  Telenor Link
> >Confidential:  no
> >Synopsis:      kaffe hangs when running javac
> >Severity:      critical
> >Priority:      high
> >Category:      kaffe
> >Class:         sw-bug
> >Release:	0.81
> >Environment:
>         System: Linux xenu 2.0.30 #6 Tue May 13 19:31:06 CEST 1997 alpha unknown
>         Architecture: alpha
>         Machine: DEC
> >Description:
> 	kaffe hangs when trying to compile something using javac.  It
> 	uses 90% CPU, but is never done.
> >How-To-Repeat:
> 	After getting kaffe to compile on Linux Alpha (had to add
> 	-Wl,-rpath,/usr/local/lib to the linking of kaffe), I tried
> 	running javac to compile test/general/RandomTest.java.  I had
> 	to add CLASSPATH=/usr/local/share/kaffe/classes.zip and
> 	LD_LIBRARY_PATH=/usr/local/lib to environment to get it to run
> 	at all.  The compiler never finishes.
> >Fix:
> 	Don't know.
> 

-- 
					-Otto.

From kaffe@kaffe.org  Thu Oct 30 20:20:50 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Thu, 30 Oct 1997 20:20:50 GMT
Subject: Finally...help with
Message-ID: <199710302020.UAA13895@home.virtual-pc.com>

building a great Web Site at an unbelievably low price.
from: sitehelp@worldwidepromo.com

Many people would like to put their business on the Web but are
frustrated trying to do it themselves or getting someone else to do it
all at a reasonable price. 

If you want to do it yourself and just need a road map for how it
is done. We can help.

If you have been working on a site and it isn't coming together the way
you would have liked. We can help.

If you don't have a clue about Web Site design and don't want anything
to do with it, but want a great site. We can help.

We can show you examples of sites we have done and explain how we did
them or we can take your idea and put it on the Web without anything
from you but input.

We offer custom sites with high quality graphics from our extensive
graphics libraries. Together with copy writing and design capabilities
we combine the use of the latest advances in the media to produce sites
that are interesting, informative and captivating. 

Please reply at:        sitehelp@worldwidepromo.com
..so we can help you get the job done.







From kaffe@kaffe.org  Thu Oct 30 13:32:25 1997
From: kaffe@kaffe.org (spinny)
Date: Thu, 30 Oct 1997 13:32:25 +0000 (GMT)
Subject: Possible licensing issue with JavaSoft?
In-Reply-To: <Pine.LNX.3.95.971030084932.11373D-100000@studmuffin.ganymede.com>
Message-ID: <Pine.LNX.3.96.971030132743.3922A-100000@borovnia.spinny.com>



On Thu, 30 Oct 1997, David Scott wrote:

> 
> On Wed, 29 Oct 1997 benjamin@marimba.com wrote:
>  
> > I remember a tale of someone who put a Java logo on his homepage in 
> > the very early days of the language, to show his support.  But then 
> > Sun threatened to sue him...  I think that Sun's legal department has 
> > been quite stupidly aggressive in the past, and is probably still so.  
> > Whether or not they stand a chance of winning, they might try.  
> > Perhaps they were just bored back then, in which case we're all safe 
> > for now ;)
> 
> My take on this issue is this: Sun is being very careful in defining its
> legal position with regard to Java(tm) because of Microsoft, whom Sun
> doesn't want mucking around with Java the way they muck around with
> everything else. The thing is, for them to defend their turf against the
> Evil Empire, they have to enforce some silly-looking boilerplate even
> among more friendly folk. 
> 
> For example, it appears that O'Reilly Associates has recently heard from
> Sun's lawyers about using the Java name without the (tm) on the covers of
> their excellent Java books. This is particularly bizarre when you consider
> how Java-friendly ORA really is. 
> 
> For the time being, I think Sun should be given very wide berth on this
> issue, until things sort themselves out a bit.  I find their lawsuit
> against Microsoft very interesting.  One way to support them is simply to
> play along. 
> 
> David Scott
> Montreal
> 

I couldn't agree more.  It would be awful if they sat around and judged
the individual intentions of all the various organizations that are in
some way involved in Java development and then, based on these
calculations, decided which organizations to prosecute.  A unified
structure works much better.  And this is a crucial juncture in the life
of Java. I do wonder how things might be different without the spectre of
Microsoft lurking about. 

garner

From kaffe@kaffe.org  Thu Oct 30 23:18:44 1997
From: kaffe@kaffe.org (Gregor Hoffleit)
Date: Fri, 31 Oct 97 00:18:44 +0100
Subject: kaffe for NEXTSTEP, was: Re: NetBSD-m68k-0.9.1 compile problem
In-Reply-To: <yam7238.1178.124108720@faui01.informatik.uni-erlangen.de>
References: <yam7238.1178.124108720@faui01.informatik.uni-erlangen.de>
Message-ID: <9710302318.AA21365@thefly.mathi.uni-heidelberg.de>

--NeXT-Mail-1425441485-1
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: quoted-printable

Matthias Hopf wrote:
    > Don't know why I stand in NOTES/HELPERS as NeXT porter [Hier war =
urspr=FCnglich eine Anlage beigef=FCgt]
   =20
    This is still correct, though. I thought I should send a message to Tim
    after my final exams (tuesday), but then, I can do it immedeately...
   =20
    Tim, could you please remove me as NeXT porter in NOTES/HELPERS? I'm
    repeatedly getting questions about the NeXT port, and I can't help those
    folks. I don't know, who's done the NeXT port, either...

I browsed my archives and found the following:

- The initial port to NEXTSTEP on m68k was done by Ralf E. Stranzenbach =
<ralf@reswi.ruhr.de> based on Matthias' Amiga code
- The port to NEXTSTEP on i386 and sparc was done by me (Gregor Hoffleit =
<flight@mathi.uni-heidelberg.de>) based on resp. code for FreeBSD/Solaris =
and with help and bugfixes from numerous other people.

Since I seem to be the only one actively working on kaffe for NEXTSTEP =
nowadays (speak up please!), Tim can feel free to add my name as "Project =
leader" for Kaffe on NEXTSTEP, i.e. for NOTES/HELPERS:

Gregor Hoffleit <flight@mathi.uni-heidelberg.de>    NEXTSTEP package of =
Kaffe


Furthermore, KAFFE 0.9.x DOESN'T WORK ON ANY NEXTSTEP platform, it's broken. =
You might want to add that to NOTES/BUGS or whereever for the next release.

	Gregor


---
|  Gregor Hoffleit     admin MATHInet / contact HeidelNeXT |
| MAIL: Mathematisches Institut   PHONE: (49)6221 56-5771 |
|       INF 288, 69120 Heidelberg / Germany  FAX: 56-3812 |
| EMAIL: flight@mathi.uni-heidelberg.de (NeXTmail)        |

--NeXT-Mail-1425441485-1
Content-Type: text/enriched; charset=iso-8859-1
Content-Transfer-Encoding: quoted-printable

Matthias Hopf wrote:

    > Don't know why I stand in NOTES/HELPERS as NeXT porter <

   =20

    This is still correct, though. I thought I should send a message to Tim

    after my final exams (tuesday), but then, I can do it immedeately...

   =20

    Tim, could you please remove me as NeXT porter in NOTES/HELPERS? I'm

    repeatedly getting questions about the NeXT port, and I can't help those

    folks. I don't know, who's done the NeXT port, either...


I browsed my archives and found the following:


- The initial port to NEXTSTEP on m68k was done by Ralf E. Stranzenbach =
<<ralf@reswi.ruhr.de> based on Matthias' Amiga code

- The port to NEXTSTEP on i386 and sparc was done by me (Gregor Hoffleit =
<<flight@mathi.uni-heidelberg.de>) based on resp. code for FreeBSD/Solaris =
and with help and bugfixes from numerous other people.


Since I seem to be the only one actively working on kaffe for NEXTSTEP =
nowadays (speak up please!), Tim can feel free to add my name as "Project =
leader" for Kaffe on NEXTSTEP, i.e. for NOTES/HELPERS:


Gregor Hoffleit <<flight@mathi.uni-heidelberg.de>    NEXTSTEP package of =
Kaffe



Furthermore, KAFFE 0.9.x DOESN'T WORK ON ANY NEXTSTEP platform, it's broken. =
You might want to add that to NOTES/BUGS or whereever for the next release.


	Gregor



---

|
<bigger>=20
</bigger><smaller> Gregor Hoffleit     admin MATHInet / contact HeidelNeXT |

| MAIL: Mathematisches Institut   PHONE: (49)6221 56-5771 |

|       INF 288, 69120 Heidelberg / Germany  FAX: 56-3812 |

| EMAIL: flight@mathi.uni-heidelberg.de (NeXTmail)        |

</smaller>

--NeXT-Mail-1425441485-1--

From kaffe@kaffe.org  Fri Oct 31 00:01:45 1997
From: kaffe@kaffe.org (Aleksey Sudakov)
Date: Thu, 30 Oct 97 19:01:45 -0500
Subject: kaffe for NEXTSTEP, was: Re: NetBSD-m68k-0.9.1 compile problem
In-Reply-To: <9710302318.AA21365@thefly.mathi.uni-heidelberg.de>
References: <yam7238.1178.124108720@faui01.informatik.uni-erlangen.de>
 <9710302318.AA21365@thefly.mathi.uni-heidelberg.de>
Message-ID: <9710310001.AA01346@rkinc.com>

<nofill>: Furthermore, KAFFE 0.9.x DOESN'T WORK ON ANY NEXTSTEP platform, it's
: broken. You might want to add that to NOTES/BUGS or whereever for the next
: release.
</nofill>

And perhaps better explain what are these trampoline functions for and why
they are not recognized by linker? Any mach assembler hacker around who
could help resolving the issue?


Aleksey

From kaffe@kaffe.org  Fri Oct 31 03:34:27 1997
From: kaffe@kaffe.org (John D. Gwinner)
Date: Thu, 30 Oct 1997 22:34:27 -0500
Subject: Kaffe bug on PC
Message-ID: <199710311521.PAA03227@home.virtual-pc.com>

Michael:

> ---"John D. Gwinner" <gwinner@northnet.org> wrote:

> > Just to cover all bases, have you examined the
> packing (#pragma pack)? 
> 
 No, under 0.9.2 the thread package does not require
> assembly. setjmp/longjmp is used.

? #pragma pack is useful even if you aren't using assembly;
I'm not sure what that has to do with setjmp/longjmp.  Maybe I'm not
following.  If you are reading 0's in structures that should have value,
this is often because the #pragma pack is wrong.  I've had problems with
this in the past.  (setting pack is a bad thing of course, but at least an
Intel machine doesn't GPF when a structure isn't aligned, but it does slow
down.).


> The problem as I see it at this point is not a union
> issue. I realized today that the problem comes from
> the initial GETSTATIC opcode processing. The
> move_ref_const macro moves the pointer
> field->info.addr to tmp which is a slots pointer then
> the macro load_ref move the contant, which is 0 (!!) to
> the stack. Later on the bytecode INVOKEVIRTUAL takes
> this stack entry as a pointer and crashes. At this 
> point I think the bug is in initializing my 
> HelloWorldApp class. This missing pointer should be 
> set when the class is loaded and processed. I hope
> that the Solaris I have next to the PC will help
> tracing the class initialization.

Hmm ... you mean the pointer is written to tmp properly, then it suddenly
becomes 0? Sounds like a wild pointer somewhere.  Set a break point on
that memory address and see what changes it to 0.  OR -- trace the macro
to make sure it's doing what it's supposed to.

I'm probably out of juice on this one, thought I could help.

		== John ==

From kaffe@kaffe.org  Fri Oct 31 18:10:18 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Fri, 31 Oct 1997 13:10:18 -0500
Subject: Important Kaffe mailing-list informations !
Message-ID: <199710311810.NAA14673@rufus.w3.org>

  Hi all,

  ====================================================
  This message has important informations about Kaffe
  mailing-list, please consider saving it for later use
  ====================================================

Remember, the mailing-list is set-up so that replying to
a message issued from the list goes back to the list.
PLEASE, take care of this !!!

  ====================================================
      the mailing list has moved, it's now served by 
               majordomo@rufus.w3.org
  ====================================================

This machine is in MIT domain, Boston, USA.
So to subscribe :

echo "subscribe" | mail majordomo@rufus.w3.org

to unsubscribe :

echo "unsubscribe" | mail majordomo@rufus.w3.org

However to send to the list use the kaffe@kaffe.org 
address, it didn't change.

  ====================================================
  Don't send unsubscribe messages to the list, use the
         majordomo@rufus.w3.org for admin message !
  ==================================================== 

 please allow a minimum amount of time for your request to
be processed, especially unsubscribing, because people always
unsubscribe from accounts different from where they are registered
(Murphy's Law applied to mailing list administration).
So I need to do it by hand !

  Considering spams and error messages forwarded to the mailing
list there is nothing (or very few things) I can do to avoid
them, please bear with the state of the art :-\ ... I just can't
fix remote broken SMTP agents. 
  If you have proven known way of improving things I would be
happy to heard from you about it.

  ======================================================
      The mailing list has an archives, read them !
        http://rufus.w3.org/tools/Kaffe/messages
      search from : http://rufus.w3.org/tools/Kaffe/
  ======================================================

  I will try to do my best to keep the quality of service to
it's maximum (I guess the delivery speed already improved by a
factor of 10), but this also need help from the people using
the list. Remember that a junk mail sent to it irritate around
600 persons !

  Thank you for having read this message, please keep it,

    yours,

  Daniel

-- 
Daniel.Veillard@w3.org | MIT/LCS   Room NE43-344 | Today's Bookmarks :
Tel : +1.617.253.5884  | 545 Technology Square   | Linux, WWW, Java,
Fax : +1.617.258.5999  | Cambridge, MA 02139 USA | badminton, Kaffe,
http://www.w3.org/People/W3Cpeople.html#Veillard | HTTP-NG and Amaya.


From kaffe@kaffe.org  Fri Oct 31 19:28:06 1997
From: kaffe@kaffe.org (Michael R Cook)
Date: 31 Oct 1997 14:28:06 -0500
Subject: Important Kaffe mailing-list informations !
In-Reply-To: Daniel.Veillard@w3.org's message of "Fri, 31 Oct 1997 13:10:18 -0500"
References: <199710311810.NAA14673@rufus.w3.org>
Message-ID: <r4yb39zq9l.fsf@cognex.com>

>>>>> "DV" == Daniel Veillard <Daniel.Veillard@w3.org> writes:

 DV> Remember, the mailing-list is set-up so that replying to
 DV> a message issued from the list goes back to the list.

Incidentally, that kind of behavior is generally considered to be
wrong: the list remailer should not be adding the Reply-to header.

In fact, the Gnus mail/news reader has special support for such
mailing lists.  The manual for Gnus says:

|`broken-reply-to'
|     Elements like `(broken-reply-to . t)' signal that `Reply-To'
|     headers in this group are to be ignored.  This can be useful if
|     you're reading a mailing list group where the listserv has inserted
|     `Reply-To' headers that point back to the listserv itself.  This is
|     broken behavior.  So there!

Michael.

From kaffe@kaffe.org  Fri Oct 31 19:46:59 1997
From: kaffe@kaffe.org (Per Bothner)
Date: Fri, 31 Oct 1997 11:46:59 -0800
Subject: Possible licensing issue with JavaSoft?
In-Reply-To: Your message of "Wed, 29 Oct 1997 16:59:18 PST."
 <199710300059.QAA20853@bach.marimba.com>
Message-ID: <199710311945.LAA06297@cygnus.com>

> It would be kind of silly for any company _not_ to want a monopoly, 
> wouldn't it?

Nope.  It's kind of silly to want a monopoly.  It invites all kinds
of unwelcome attention (legally, upstart comptetitors, potshots from
journalists).  It also tends to make you too complacent, which is
dangerous in the long run.

What you want to be is the largest supplier, the one with the
brand-name recogniztion, and that nobody gets fired for buying.
E.g. Intel and Microsoft.

	--Per Bothner
Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner


From kaffe@kaffe.org  Fri Oct 31 21:25:32 1997
From: kaffe@kaffe.org (Clyde Smith-Stubbs)
Date: Sat, 1 Nov 1997 07:25:32 +1000
Subject: Possible licensing issue with JavaSoft?
In-Reply-To: <199710311945.LAA06297@cygnus.com>; from Per Bothner on Fri, Oct 31, 1997 at 11:46:59AM -0800
References: <199710300059.QAA20853@bach.marimba.com> <199710311945.LAA06297@cygnus.com>
Message-ID: <19971101072532.17804@htsoft.com>

On Fri, Oct 31, 1997 at 11:46:59AM -0800, Per Bothner wrote:
> Nope.  It's kind of silly to want a monopoly.  It invites all kinds
> of unwelcome attention (legally, upstart comptetitors, potshots from

Having multiple vendors also grows the total market size - often
resulting in more for each vendor than they could have expected if
they had to do that all themselves.

-- 
Clyde Smith-Stubbs               |            HI-TECH Software
Email: clyde@htsoft.com          |          Phone            Fax
WWW:   http://www.htsoft.com/    | USA: (408) 490 2885  (408) 490 2885
PGP:   finger clyde@htsoft.com   | AUS: +61 7 3354 2411 +61 7 3354 2422
---------------------------------------------------------------------------
ANSI C for the PIC! Now shipping! See www.htsoft.com for more info.

From kaffe@kaffe.org  Fri Oct 31 21:41:44 1997
From: kaffe@kaffe.org (Lee Elson)
Date: Fri, 31 Oct 1997 13:41:44 -0800
Subject: Linux and biss-awt
Message-ID: <9710311341.ZM20919@magus.jpl.nasa.gov>

I'm a new user and am having a problem setting things up. The error
message I get is:
java.lang.UnsatisfiedLinkError: no bissawt in shared library path
        at java/lang/Throwable.<init>(line unknown, pc 0x836a90f)
        at java/lang/Error.<init>(line unknown, pc 0x836aa61)
        at java/lang/LinkageError.<init>(line unknown, pc 0x836a9e9)
        at java/lang/UnsatisfiedLinkError.<init>(line unknown, pc
0x836a971)
        at java/lang/Runtime.loadLibrary(line unknown, pc 0x8368b27)
        at java/lang/System.loadLibrary(line unknown, pc 0x836a721)
        at biss/awt/kernel/NativeLib.<clinit>(line unknown, pc 0x8366e16)
.
.
.
.
Here's what I have in /usr/local/lib:
libkaffe_bissawt.so.0.87  libnative.so.0.92         libzip.so.0.92
libagent.so@              libkaffevm.so@            libnet.so@
libagent.so.0.92          libkaffevm.so.0.92        libnet.so.0.92
libkaffe_bissawt.so@      libnative.so@             libzip.so@

I have LD_LIBRARY_PATH set to /usr/local/lib and I've run ldconfig. Is
libkaffe_bissawt.so.0.87 the latest version? Anyone see any other problems
which cause the above error? Any help is appreciated.

Lee Elson (elson@magus.jpl.nasa.gov)

From kaffe@kaffe.org  Wed Oct  1 14:42:59 1997
From: kaffe@kaffe.org (Mikael Ståldal)
Date: Wed, 01 Oct 1997 14:42:59 +0100
Subject: Porting Kaffe to OS/2
Message-ID: <jNlM0EpfhKYU092yn@ingen.reklam.nada.kth.se>

I'm trying to port Kaffe to OS/2 (an i386 platform) using the EMX compiler
(GCC port). I have encountered some problems.

Some different header files differ only in case, and since the file system
in OS/2 isn't case sensitive this causes clashes. String.h clashes with the
standard string.h and System.h clashes with Kaffe's system.h.

Make doesn't find machine.c in the interp subdirectory, how is it supposed
to find it there? I'm using GNU Make 3.75. When I copied it to kaffevm it
works. The same problem with the package directoris in packages.

When I copied files from the packages, things broke due to "../../.." style
#include directives that I hade to change manually. Try to avoid using ".."
in #include directives.

I get the following error:

make[1]: Entering directory `/UNIX/kaffe/kaffe/kaffevm'
gcc -g -O -Zexe -DNO_SHARED_LIBRARIES -I. -I. -I./../../config -I../../config -I../../include -I./../../include -DINTERPRETER -I./intrp   -DKVER=\"0.91\" -c  machine.c
machine.c: In function `virtualMachine':
machine.c:141: structure has no member named `exception_table_len'
machine.c:152: structure has no member named `ins'
machine.c:178: structure has no member named `code'
machine.c:182: structure has no member named `codelen'
make[1]: *** [machine.o] Error 1
make[1]: Leaving directory `/UNIX/kaffe/kaffe/kaffevm'
make: *** [all] Error 1

When I look at the structure, it has no exception_table_len member. How is
this supposed to works?

-- 
/****************************************************************\
* You have just read a message from Mikael Ståldal.              *
*                                                                *
* Remove "ingen.reklam." from the address before mail replying.  *
\****************************************************************/

From kaffe@kaffe.org  Thu Oct  2 01:06:44 1997
From: kaffe@kaffe.org (Ram Madduluri)
Date: Wed, 01 Oct 1997 18:06:44 -0600
Subject: Kaffe runtime question
Message-ID: <s4328dbf.005@novell.com>

Hello,

I have ported Kaffe 0.9.1 to OSF/1 on x86 platform. The initial
HelloWorldApp test passed. However, when I run a sample server application -
UDPEchoServer, I get the following error:
	$ kaffe UDPEchoServer
	My IP: 127.0.0.1
	My port: 6666
	Failed to locate native function:
        	java/net/PlainDatagramSocketImpl.datagramSocketCreate()V
	java.lang.UnsatisfiedLinkError
        	at java/net/DatagramSocket.create(133)
        	at java/net/DatagramSocket.<init>(114)
        	at java/net/DatagramSocket.<init>(100)
        	at UDPServer.main(41)

The PlainDatagramSocketImpl class is in the
/usr/local/share/kaffe/classes.zip file. Am I missing something here? I can
run some of the other tests just fine (DemoServer, DemoClient, ..). Is there
some configuration stuff that I messed up??

Any pointers would be greatly appreciated.

Regards,

Ram
--
Ram Madduluri                                                  E-Mail:
ram@Novell.COM
Corporate Research                                          Voice:  
801-222-5361
Novell Inc.                                                          Fax:   
  801-222-5177
1555 North Technology Way
Orem, UT 84097

From kaffe@kaffe.org  Thu Oct  2 05:33:54 1997
From: kaffe@kaffe.org (Orville R. Weyrich)
Date: Wed, 1 Oct 1997 21:33:54 -0700 (MST)
Subject: Porting Kaffe to OS/2
In-Reply-To: <jNlM0EpfhKYU092yn@ingen.reklam.nada.kth.se>
Message-ID: <Pine.LNX.3.91.971001210131.6624A-100000@dopey.weyrich.com>

On Wed, 1 Oct 1997, Mikael Ståldal wrote:

> I'm trying to port Kaffe to OS/2 (an i386 platform) using the EMX compiler
> (GCC port). I have encountered some problems.
> 
> Some different header files differ only in case, and since the file system
> in OS/2 isn't case sensitive this causes clashes. String.h clashes with the
> standard string.h and System.h clashes with Kaffe's system.h.

My suggestion is to place all kaffe include files in a directory named
kaffe, such that the kaffe directory is in your INCLUDE path.  This means
that you will have to modify the source code to:

	#include <string.h>

and

	#include <kaffe/String.h>

It is less work if you only do this for the files which conflict, but it
is cleaner if you move all the kaffe headers to this location and tweak the
source code appropriately.  In my opinion, this would add value to the 
source code base for all users, since it would make clear which includes
are for system headers, and which are for kaffe headers.

I just downloaded kaffe myself, and haven't looked at the code yet,
so some other variation may be more consistent with the existing
organization, but this should get you started.  I use OS/2 and would
love to have kaffe on OS/2, so give me a shout if I can be of any
more help :-)

> 
> Make doesn't find machine.c in the interp subdirectory, how is it supposed
> to find it there? I'm using GNU Make 3.75. When I copied it to kaffevm it
> works. The same problem with the package directoris in packages.

Is there a configuration script that generates machine.c based on the
environment?
 
> When I copied files from the packages, things broke due to "../../.." style
> #include directives that I hade to change manually. Try to avoid using ".."
> in #include directives.

I am afraid that there is no general solution for this portability problem,
absolute paths also have their problems.  Perhaps a perl script containing
something like

	if ( $line =~ /^#include/ )
	{
		$line =~ s/\.\.\//..\\/;
	}

might be the easiest to do over all your source files.
(i.e. find all include lines and change ../ to ..\).

> 
> I get the following error:
> 
> make[1]: Entering directory `/UNIX/kaffe/kaffe/kaffevm'
> gcc -g -O -Zexe -DNO_SHARED_LIBRARIES -I. -I. -I./../../config 
-I../../config -I../../include -I./../../include -DINTERPRETER -I./intrp   -DKVER=\"0.91\" -c  machine.c
> machine.c: In function `virtualMachine':
> machine.c:141: structure has no member named `exception_table_len'
> machine.c:152: structure has no member named `ins'
> machine.c:178: structure has no member named `code'
> machine.c:182: structure has no member named `codelen'
> make[1]: *** [machine.o] Error 1
> make[1]: Leaving directory `/UNIX/kaffe/kaffe/kaffevm'
> make: *** [all] Error 1
> 
> When I look at the structure, it has no exception_table_len member. How is
> this supposed to works?
> 

I don't know.  It may be due to using the wrong header file.

===================================================================
Orville R. Weyrich, Jr.                 Weyrich Computer Consulting
mailto:orville@weyrich.com                   http://www.weyrich.com
-------------------------------------------------------------------

From kaffe@kaffe.org  Thu Oct  2 13:14:41 1997
From: kaffe@kaffe.org (David Turner)
Date: Thu, 2 Oct 1997 13:14:41 +0100 (MET DST)
Subject: Porting Kaffe to OS/2
In-Reply-To: <jNlM0EpfhKYU092yn@ingen.reklam.nada.kth.se> from "Mikael Ståldal" at Oct 1, 97 02:42:59 pm
Message-ID: <199710021114.NAA08384@messiaen.enst.fr>

Hi Mikael

Mikael Ståldal said:
> 
> machine.c: In function `virtualMachine':
> machine.c:141: structure has no member named `exception_table_len'
> machine.c:152: structure has no member named `ins'
> machine.c:178: structure has no member named `code'
> machine.c:182: structure has no member named `codelen'
> make[1]: *** [machine.o] Error 1
> make[1]: Leaving directory `/UNIX/kaffe/kaffe/kaffevm'
> make: *** [all] Error 1
> 
> When I look at the structure, it has no exception_table_len member. How is
> this supposed to works?
>

It's a famous bug in the 0.9.1 release (it's not EMX specific ;-)
The fix has been released in the mail archive, but I'll give it here
again, from memory :

  1. replace 'method->codelen' by 'method->c.bcode.codelen'  (line 182)
     and     'method->code'    by 'method->c.bcode.code'     (line 178)

     ( I don't remember if it's "method" or "meth" or "m", but it's
       a pointer to a method structure )

  2. replace 'exception_table_len > 0' by 'exception_table != 0' (line 141)

  3. define a local variable "char* str" and replace
     line 152 with :

      "str = meth->signature->data;
       idx = sizeofSig( &str, false ) + ( .... & ..... ? ... : ... )'
                                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                                          same as before..

  Then recompile, it should work OK now.. ( well, there are still some
  bugs in there ;-)

Good luck, you'll need it..

- David

From kaffe@kaffe.org  Thu Oct  2 17:59:03 1997
From: kaffe@kaffe.org (Tony Kimball)
Date: Thu, 2 Oct 1997 11:59:03 -0500 (CDT)
Subject: Porting Kaffe to OS/2
References: <jNlM0EpfhKYU092yn@ingen.reklam.nada.kth.se>
 <Pine.LNX.3.91.971001210131.6624A-100000@dopey.weyrich.com>
Message-ID: <199710021659.LAA06487@compound.east.sun.com>

Quoth Orville R. Weyrich on Wed, 1 October:
:  
: > When I copied files from the packages, things broke due to "../../.." style
: > #include directives that I hade to change manually. Try to avoid using ".."
: > in #include directives.
: 
: I am afraid that there is no general solution for this portability problem,
: absolute paths also have their problems. 

-I fixes it everywhere.

From kaffe@kaffe.org  Thu Oct  2 18:49:10 1997
From: kaffe@kaffe.org (James A. Hillyerd)
Date: Thu, 2 Oct 1997 10:49:10 -0700 (PDT)
Subject: Next Release?
Message-ID: <Pine.LNX.3.95.971002104801.6765B-100000@animal.blarg.net>

Any guesses on when the next release of Kaffe is coming out?  This list
has been pretty quiet lately. =)

-james

[  James A. Hillyerd (JH2162) - james@blarg.net - Web Developer  ]
[  http://www.blarg.net/~james/   http://www.hyperglyphics.com/  ]
[ 1024/B11C3751 CA 1C B3 A9 07 2F 57 C9  91 F4 73 F2 19 A4 C5 88 ]

From kaffe@kaffe.org  Sat Oct  4 04:49:11 1997
From: kaffe@kaffe.org (Michael Hirsch)
Date: Fri, 3 Oct 1997 23:49:11 -0400 (EDT)
Subject: java.lang.ClassNotFoundException:HelloWorldApp
Message-ID: <199710040349.XAA22083@cssun.mathcs.emory.edu>

I have the same problem that Bryan Felix reported.

I tried to follow the README instructions exactly.  I got all four
tarballs: 
kaffe-0.9.1.tgz 
kaffe-0.9.1-package-javasoft.com.tgz
kaffe-0.9.1-package-biss-net.com.tgz and
kaffe-0.9.1-package-epfl.ch.tgz 
and unpacked them in the same directory.

./configure worked fine, as did make and make install.

But now:
[34] test> echo $CLASSPATH $KAFFEHOME
.:/usr/local/share/kaffe/classes.zip:/usr/local/share/kaffe/biss.zip /usr/local/share/kaffe
[35] test> pwd
/usr/local/Uploads/kaffe-0.9.1/test
[36] test> kaffe HelloWorldApp
java.lang.ClassNotFoundException: HelloWorldApp
[37] test> 

I tried to do exactly as it says.  . is in the CLASSPATH.
I did not compile HelloWorldApp.fava into a .class file because 1) I
don't know how, exactly, and 2) the README doesn't say to.

However, if I try
[37] test> javac HelloWorldApp.java
my screen fills with this error message:
java.lang.NullPointerException
	at java/io/File.list(line unknown, pc 0x8520a6c)
	at sun/tools/java/ClassPathEntry.getFiles(line unknown, pc 0x8532197)
	at sun/tools/java/ClassPath.getFile(line unknown, pc 0x852ddb9)
	at sun/tools/java/ClassPath.getFile(line unknown, pc 0x8530bd6)
	at sun/tools/java/Package.getBinaryFile(line unknown, pc 0x853477a)
	at sun/tools/java/Package.classExists(line unknown, pc 0x852083b)
	at sun/tools/javac/BatchEnvironment.classExists(line unknown, pc 0x852c650)
	at sun/tools/java/Environment.classExists(line unknown, pc 0x8530369)
	at sun/tools/java/Environment.classExists(line unknown, pc 0x8530369)
	at sun/tools/java/Imports.importable(line unknown, pc 0x851ee55)
	at sun/tools/java/Imports.resolve(line unknown, pc 0x852d2db)
	at sun/tools/java/ImportEnvironment.resolve(line unknown, pc 0x8521eb3)
	at sun/tools/java/Environment.resolve(line unknown, pc 0x85205a2)
	at sun/tools/java/Environment.resolveName(line unknown, pc 0x851dea4)
	at sun/tools/java/Environment.resolveName(line unknown, pc 0x851dcca)
	at sun/tools/javac/SourceClass.resolveSuper(line unknown, pc 0x851ecd4)
	at sun/tools/javac/SourceClass.resolveSupers(line unknown, pc 0x852e895)
	at sun/tools/javac/SourceClass.resolveTypeStructure(line unknown, pc 0x852e1bd)
	at sun/tools/javac/SourceClass.basicCheck(line unknown, pc 0x851f971)
	at sun/tools/java/ClassDeclaration.getClassDefinition(line unknown, pc 0x851eb11)
	at sun/tools/javac/Main.compile(line unknown, pc 0x81b9992)
	at sun/tools/javac/Main.main(line unknown, pc 0x81a1d83)
error: An exception has occurred in the compiler; please file a bug report (http://java.sun.com/cgi-bin/bugreport.cgi).
1 error

so clearly something is wrong with my setup.  I'm sure I'm missing
something obvious.

I tried 'make test', but each test failed with what appears to be
essentially the same error as the one quoted above.

Thanks in advance,

-- 
Michael D, Hirsch			Work: (404) 727-7940 
Emory University, Atlanta, GA 30322	FAX: (404) 727-5611
email:  hirsch@mathcs.emory.edu		http://www.mathcs.emory.edu/~hirsch/

Public key for encrypted mail available upon request (or finger
hirsch@cssun.mathcs.emory.edu).

From kaffe@kaffe.org  Sun Oct  5 06:57:09 1997
From: kaffe@kaffe.org (Henrik Martin)
Date: Sat, 4 Oct 1997 22:57:09 -0700 (PDT)
Subject: Working version of kaffe for BSDI 3.0?
Message-ID: <199710050557.WAA29595@u2.netgate.net>


Hi folks! I'm trying to build kaffe v0.9.1 on a x86 machine running
BSDI 3.0, but it won't build. After having changed the "global" to "globl"
in the config/i386/trampolines.c file, I get the following linking error:

make -k
rm -f libkaffe_vm.a
ln -s libkaffe_vm.a.0.91 libkaffe_vm.a
gcc -g -O  -I. -I./../kaffevm -I../../config -I./../../config -I../../include -I./../../include -o kaffe main.o -L../kaffevm -lkaffe_vm  -L../../packages/tjwassoc.co.uk/APIcore/lib -lkaffe_native -L../../packages/tjwassoc.co.uk/APInet/lib -lkaffe_net -L../../packages/tjwassoc.co.uk/APIzip/lib -lkaffe_zip  -lipc -L/usr/X11R6/lib -lX11 -lkaffe_vm -lipc -lm -lc
ld: premature end of file in symbols of ../kaffevm/libkaffe_vm.a(machine.o)
*** Error code 1 (continuing)
`all' not remade because of errors.

I do have the kaffe-0.9.1-package-javasoft.com.tgz stuff installed
BTW.  Any hints on what may be wrong? If anyone has built any recent
version of kaffe successfully on BSDI 3.0, please let me know.

If you reply to the mailing list, please also cc: henrik@netgate.net.

Thanks,

/Henrik

From kaffe-announce@kaffe.org  Sun Oct  5 15:22:07 1997
From: kaffe-announce@kaffe.org (Tim Wilkinson)
Date: Sun, 05 Oct 1997 15:22:07 +0100
Subject: [FINALLY] Kaffe 0.9.2 released.
Message-ID: <3437A28F.41C67EA6@tjwassoc.co.uk>

KAFFE v0.9.2 - A JIT and interpreting virtual machine to run Java(tm)*
code
===========================================================================

                *** Kaffe is now JDK 1.1.2 compliant ***

Kaffe is a virtual machine design to execute Java bytecode.
This machine can be configured in two modes.  In one mode it operates as
a pure bytecode interpreter (not unlike Javasoft's machine); in the
second
mode if performs "just-in-time" code conversion from the abstract code
to
the host machine's native code.  This will ultimately allow execution of
Java code at the same speed as standard compiled code but while
maintaining
the advantages and flexibility of code independence.

New Features
------------

This version of Kaffe has the following new features:

 * Kaffe is now JDK 1.1.3 compliant !

 * Improved memory efficiency when allocating object.

 * Improvements to GC.
 
 * Alpha port partially fixed ... still lots of problems though. 
 
 * Lots of bug fixes (see Changelog)

Platforms
---------

Kaffe will run on the following platforms ('J' indicates 'just-in-time'
support, 'I' indicates 'interpreter' support):

 
========================================================================
  uProc     | Mode |    Systems
 
========================================================================
  i386      | J+I  |    FreeBSD 2.x +   Unixware        NetBSD 1.x
            |      |    Solaris 2.x     BSDI 2.x        Linux 2.0.0 +
            |      |    SCO 3.2v5       NeXTStep 3.x    Windows '95
            |      |    DG/UX           OpenBSD 2.x     OpenStep 4.x
            |      |    QNX 4.x
 
------------------------------------------------------------------------
  Sparc     | J+I  |    SunOS 4.x       Solaris 2.x     NetBSD 1.x
            |      |    NeXTStep 3.x    Linux           Fujitsu UXP/DS
            |      |    BSDI 3.x
 
------------------------------------------------------------------------
  Alpha     | J+I  |    Linux           OSF/1 
 
------------------------------------------------------------------------
  M68K      | J+I  |    AmigaOS         NeXTStep 3.x    NetBSD 1.x
            |      |    SunOS 4.x       Linux           AUX
            |      |    OpenBSD 2.x
 
------------------------------------------------------------------------
  PowerPC   | I    |    MkLinux         MachTen 4.0.3   AIX
 
------------------------------------------------------------------------
  MIPS      | I    |    IRIX 5 & 6      NetBSD 1.x
 
------------------------------------------------------------------------
  PARISC    | I    |    HPUX 10.x
 
------------------------------------------------------------------------  
  StrongARM | I    |    RiscIX          (work-in-progress)
 
========================================================================

I encourage people to make ports to other platforms and to feed them
back
to me.  I would recommend that initial ports concentrate on the
interpreter.
These can then be used as the basis for developing JIT versions for
these
systems.

The sources for Kaffe can be found at the following location:

Europe:
        ftp://ftp.kaffe.org/pub/kaffe/
        ftp://ftp.sarc.city.ac.uk/pub/kaffe/
        ftp://ftp.doc.ic.ac.uk/packages/kaffe/
        ftp://ftp.lysator.liu.se/pub/languages/java/kaffe/
        ftp://sunsite.mff.cuni.cz/Languages/Java/kaffe/
        ftp://sunsite.auc.dk/pub/languages/java/kaffe/
        ftp://alterego.stack.urc.tue.nl/pub/kaffe/
        ftp://ftp.fh-wolfenbuettel.de/pub/lang/java/kaffe/
        ftp://ftp.uni-erlangen.de/pub/source/Java/kaffe/
        ftp://gd.tuwien.ac.at/languages/java/kaffe/
USA:
        ftp://ftp.cs.columbia.edu/pub/kaffe/
        ftp://sunsite.unc.edu/pub/languages/java/kaffe/
Japan:
        ftp://ftp.webcity.co.jp/pub/andoh/java/kaffe/
Others:
        ftp://ftp.unicamp.br/pub/languages/java/kaffe/


What you need
-------------

For a Kaffe system you need the following parts:

 kaffe-0.9.2.tgz                        Virtual machine and all bits
                                        provided by tjwassoc.co.uk
 kaffe-0.9.2-package-javasoft.com.tgz   Sun's class libraries.

These two are are the minimum to get something useful working (you might
already have the classes libraries from JDK 1.1.2; if so don't bother to
get them again).

If you want to use graphics you'll need one of the following:

 kaffe-0.9.2-package-biss-net.com.tgz   AWT graphics support.

Uncompress and untar these archives into the same directory, configure,
make and install.

Enjoy.

Tim Wilkinson
<tim@kaffe.org>
----
* Java is a registered trademark of Sun Microsystems, Inc.

--
  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
  London, UK.                           Email:   tim@tjwassoc.co.uk

From kaffe@kaffe.org  Sun Oct  5 17:00:14 1997
From: kaffe@kaffe.org (Matthias Hopf)
Date: Sun, 5 Oct 1997 18:00:14 +0200 (MET DST)
Subject: Next Release?
In-Reply-To: <Pine.LNX.3.95.971002104801.6765B-100000@animal.blarg.net> from "James A. Hillyerd" at Oct 2, 97 10:49:10 am
Message-ID: <199710051600.SAA26186@faui01.informatik.uni-erlangen.de>

> Any guesses on when the next release of Kaffe is coming out?  This list
> has been pretty quiet lately. =)
> 
> -james

Kaffe v0.9.2 has just been released. But as I have to learn for final
exams right now, I won't be able to do anything about kaffe until
end of october. So expect a tested version of V0.9.2 (or V0.9.3 then ;)
in november.

And yes, it has been quiet here the last months. Nothing happened about
kaffe either, lots of the developers seemed to be on holiday ;)
I didn't do anything about kaffe0.9.1 as it seemed to have several
unfixed bugs.

CU

Matthias

-- 
    //             |   Matthias Hopf - "Hoeppel"   |    _      __
\\ //    Amiga     |  student of computer science  |  _|cience |-iction
 \X/ by conviction |      in Erlangen/Germany      | by belief in Future

 EMail:  mshopf@informatik.uni-erlangen.de
 Aminet: ftpamiga@epix.rrze.uni-erlangen.de
 WWW:    http://wwwcip.informatik.uni-erlangen.de/user/mshopf/

From kaffe@kaffe.org  Sun Oct  5 17:59:23 1997
From: kaffe@kaffe.org (Mark Huizer)
Date: Sun, 5 Oct 1997 18:59:23 +0200
Subject: [FINALLY] Kaffe 0.9.2 released.
In-Reply-To: <3437A28F.41C67EA6@tjwassoc.co.uk>; from "Tim Wilkinson" on Sun, Oct 05, 1997 at 03:22:07PM +0100
References: <3437A28F.41C67EA6@tjwassoc.co.uk>
Message-ID: <19971005185923.36519@xaa.stack.nl>

> 
>                 *** Kaffe is now JDK 1.1.2 compliant ***
> 
>  * Kaffe is now JDK 1.1.3 compliant !

Nice :)

What's the status of the BISS AWT classes by now? I missed out a few
weeks, so there's the lack of knowledge. Is it possible to run graphical
stuff a bit better by now?
-------------------------------------------------------------------------
- Mark Huizer  - xaa@stack.nl     - rcbamh@urc.tue.nl                   -
-------------------------------------------------------------------------
- Anyone who is capable of getting themselves made President should	-
- on no account be allowed to do the job (The Hitchhikers Guide)	-
-------------------------------------------------------------------------

From kaffe@kaffe.org  Sun Oct  5 19:53:36 1997
From: kaffe@kaffe.org (Bryan Felix)
Date: Sun, 5 Oct 1997 11:53:36 -0700 (MST)
Subject: Working version of kaffe for BSDI 3.0?
Message-ID: <199710051853.LAA20460@rapidaxcess.com>

>Hi folks! I'm trying to build kaffe v0.9.1 on a x86 machine running
>BSDI 3.0, but it won't build. After having changed the "global" to "globl"
>in the config/i386/trampolines.c file, I get the following linking error:
>
>If you reply to the mailing list, please also cc: henrik@netgate.net.
>
>Thanks,
>
>/Henrik
>
        I have done my best to get any version running on BSDI 3.0 and on
BSDI 2.1 and finally cut it off. Only so much time to spare on one project.
I decided this project is just to buggy and am waiting patiently for a
stable developement to be released.
        I did manage to get version 0.9.0 to the point that it would at
least just hang after I went through typing in the CLASSPATH by hand with
each execution. I was in touch with another user on this list that has
gotten an even older version to run on BSDI though he has to add the
CLASSPATH on to every execution as well.
        Sorry I can't be of much more help, but thought I would add my
experience for what it's worth.

With Regret,
Bryan Felix


                            _______________________________________________
        $$$$$$$$$$$$$$$$               ---Rapid Axcess---
          $$$$$$$$$$$$$$$$$    ...Premium Web Design & Domain Hosting...
           $$$$     $$$$$$$$     ______________________________________ 
           $$$$      $$$$$$$    (_____       Bryan Felix        _______)
           $$$      $$$$$   __________) felix@rapidaxcess.com  (__
          $$$     $$$$$    (_______      Voice: 602-952-8275  ______)
         $$$$  $$$$$           ____)     Page: 602-201-4646  (____________
        $$$$$$$$$             (__________ Fax: 602-224-9585 ______________)
       $$$   $$$$        $$$        
      $$$     $$$$      $$ $$       PO Box 61413 
    $$$        $$$$$   $$___$$      Phoenix, AZ 85082-1413
 $$$$$$        $$$$$$$$$     $$ ____http://www.rapidaxcess.com_____________

From kaffe@kaffe.org  Sun Oct  5 22:32:10 1997
From: kaffe@kaffe.org (Michael Hirsch)
Date: Sun, 5 Oct 1997 17:32:10 -0400 (EDT)
Subject: java.lang.ClassNotFoundException:HelloWorldApp
In-Reply-To: <199710040349.XAA22083@cssun.mathcs.emory.edu>
References: <199710040349.XAA22083@cssun.mathcs.emory.edu>
Message-ID: <199710052132.RAA23133@cssun.mathcs.emory.edu>

Okay, I found some of the problem.  I wrote:

>>>>> In article <199710040349.XAA22083@cssun.mathcs.emory.edu>, Michael Hirsch <hirsch@mathcs.emory.edu> writes:

Michael> I have the same problem that Bryan Felix reported.  I tried
Michael> to follow the README instructions exactly.  I got all four
Michael> tarballs: kaffe-0.9.1.tgz
Michael> kaffe-0.9.1-package-javasoft.com.tgz
Michael> kaffe-0.9.1-package-biss-net.com.tgz and
Michael> kaffe-0.9.1-package-epfl.ch.tgz and unpacked them in the same
Michael> directory.

Michael> ./configure worked fine, as did make and make install.

Michael> But now: 
Michael> [34] test> echo $CLASSPATH $KAFFEHOME
Michael>
.:/usr/local/share/kaffe/classes.zip:/usr/local/share/kaffe/biss.zip /usr/local/share/kaffe 
MIchael> [35] test> pwd
Michael> /usr/local/Uploads/kaffe-0.9.1/test 
Michael> [36] test> kaffe HelloWorldApp 
Michael> java.lang.ClassNotFoundException: HelloWorldApp

This, of course, does not work because the .java file has not been
compiled into a .class file.  Unfortunately, the README file says that
this should work.  Perhaps the next release will add the compiled
HelloWorldApp.class file to the test directory so silly people like me
don't get confused.

I made a compiled the .java file on another system and then tried to
run it and it worked fine.  So the interpreter/JIT seems to be okay.

Next, I tried compiling and that is still broken.  I reinstalled
everything and I still get these errors.  Any further suggestions?
It kind of looks like I don't have something installed, but I
installed all 4 .tgz files.

Thanks,

--Michael

Here are the errors:
[2] test> javac HelloWorldApp.java 
java.lang.NullPointerException
	at java/io/File.list(line unknown, pc 0x8520a6c)
	at sun/tools/java/ClassPathEntry.getFiles(line unknown, pc 0x8532197)
	at sun/tools/java/ClassPath.getFile(line unknown, pc 0x852ddb9)
	at sun/tools/java/ClassPath.getFile(line unknown, pc 0x8530bd6)
	at sun/tools/java/Package.getBinaryFile(line unknown, pc 0x853477a)
	at sun/tools/java/Package.classExists(line unknown, pc 0x852083b)
	at sun/tools/javac/BatchEnvironment.classExists(line unknown, pc 0x852c650)
	at sun/tools/java/Environment.classExists(line unknown, pc 0x8530369)
	at sun/tools/java/Environment.classExists(line unknown, pc 0x8530369)
	at sun/tools/java/Imports.importable(line unknown, pc 0x851ee55)
	at sun/tools/java/Imports.resolve(line unknown, pc 0x852d2db)
	at sun/tools/java/ImportEnvironment.resolve(line unknown, pc 0x8521eb3)
	at sun/tools/java/Environment.resolve(line unknown, pc 0x85205a2)
	at sun/tools/java/Environment.resolveName(line unknown, pc 0x851dea4)
	at sun/tools/java/Environment.resolveName(line unknown, pc 0x851dcca)
	at sun/tools/javac/SourceClass.resolveSuper(line unknown, pc 0x851ecd4)
	at sun/tools/javac/SourceClass.resolveSupers(line unknown, pc 0x852e895)
	at sun/tools/javac/SourceClass.resolveTypeStructure(line unknown, pc 0x852e1bd)
	at sun/tools/javac/SourceClass.basicCheck(line unknown, pc 0x851f971)
	at sun/tools/java/ClassDeclaration.getClassDefinition(line unknown, pc 0x851eb11)
	at sun/tools/javac/Main.compile(line unknown, pc 0x81b9992)
	at sun/tools/javac/Main.main(line unknown, pc 0x81a1d83)
error: An exception has occurred in the compiler; please file a bug report (http://java.sun.com/cgi-bin/bugreport.cgi).
1 error
[3] test> 

From kaffe@kaffe.org  Mon Oct  6 18:21:09 1997
From: kaffe@kaffe.org (P. V. Thatayya Naidu)
Date: Mon, 6 Oct 1997 10:21:09 -0700
Subject: Object Seralistion!
Message-ID: <199710061721.KAA01323@faith>

Hi,
    Will  Object Serialistion and RMI work with this new release Kaffe 0.9.2 ?
    
    
thanks
naidu
P. V. Thatayya Naidu

From kaffe@kaffe.org  Mon Oct  6 14:03:55 1997
From: kaffe@kaffe.org (Tim Wilkinson)
Date: Mon, 06 Oct 1997 14:03:55 +0100
Subject: BISS-AWT and Kaffe 0.9.2
Message-ID: <3438E1BB.1CFBAE39@tjwassoc.co.uk>

All,

Please note that I've withdrawn the BISS-AWT and EPFH.CH packages from
the latest release since neither of them work at the moment.  Hopefull
BISS-AWT will be back soon.

Regards
Tim
--
  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
  London, UK.                           Email:   tim@tjwassoc.co.uk

From kaffe@kaffe.org  Mon Oct  6 14:45:46 1997
From: kaffe@kaffe.org (Peter C. Mehlitz)
Date: Mon, 6 Oct 1997 15:45:46 +0200
Subject: [FINALLY] Kaffe 0.9.2 released.- BISS-AWT status
In-Reply-To: Mark Huizer <xaa@stack.nl>
 "Re: [FINALLY] Kaffe 0.9.2 released." (Oct  5,  6:59pm)
References: <3437A28F.41C67EA6@tjwassoc.co.uk>
 <19971005185923.36519@xaa.stack.nl>
Message-ID: <9710061545.ZM280@mwave.biss-net.com>

> Is it possible to run graphical stuff a bit better by now?

Sorry, not yet released. We have the biss.jde up ad running again, but before
we go public, this time we want to improve java.awt compatibility. This is even
more important since the java.awt classes are replaced as well. We also still
need the go from our sponsor.

Peter

From kaffe@kaffe.org  Mon Oct  6 17:25:54 1997
From: kaffe@kaffe.org (TIOOYK)
Date: Mon, 6 Oct 1997 09:25:54 -0700
Subject: Working version of kaffe for BSDI 3.0?
References: <199710051853.LAA20460@rapidaxcess.com>
Message-ID: <199710061625.JAA00292@van-as-05b01.direct.ca>

On Sun, Oct 05, 1997 at 11:53:36AM -0700, Bryan Felix wrote:
: / >Hi folks! I'm trying to build kaffe v0.9.1 on a x86 machine running
: / >BSDI 3.0, but it won't build. After having changed the "global" to "globl"
: / >in the config/i386/trampolines.c file, I get the following linking error:
: / >
: / >If you reply to the mailing list, please also cc: henrik@netgate.net.
: / >
: / >Thanks,
: / >
: / >/Henrik
: / >
: /         I have done my best to get any version running on BSDI 3.0 and on
: / BSDI 2.1 and finally cut it off. Only so much time to spare on one project.
: / I decided this project is just to buggy and am waiting patiently for a
: / stable developement to be released.
: /         I did manage to get version 0.9.0 to the point that it would at
: / least just hang after I went through typing in the CLASSPATH by hand with
: / each execution. I was in touch with another user on this list that has
: / gotten an even older version to run on BSDI though he has to add the
: / CLASSPATH on to every execution as well.
: /         Sorry I can't be of much more help, but thought I would add my
: / experience for what it's worth.
: / 

What's the problem? I use BSD/OS 2.1 and running kaffe -version I get:

# kaffe -version
Kaffe Virtual Machine
Copyright (c) 1997 T. J. Wilkinson & Associates, London, UK.
Engine: Just-in-time   Version: 0.91   Java Version: 1.1.2

I don't recall doing much to get things running, apart from the .globl
thing.

Mind you, there is so much that DOESN'T work past the tests that it's
not worth compiling right now : <

To be fair it's the graphic stuff that is pretty broken..  Perhaps
0.9.2 has a useable companion AWT library now??  Guess I'll find out
soon enough.

-g

--
Berbeee-a seys,
 Teke-a qooeeloodes in geen und gu tu a deescu reeght evey!
 Bork Bork Bork!
 Boot Kee seys, VOO-VOO!!  Nu credeet et ``Mr. Leeqoour''!!

From kaffe@kaffe.org  Mon Oct  6 23:34:36 1997
From: kaffe@kaffe.org (Tim Wilkinson)
Date: Mon, 06 Oct 1997 23:34:36 +0100
Subject: Object Seralistion!
References: <199710061721.KAA01323@faith>
Message-ID: <3439677C.3F54BC7E@tjwassoc.co.uk>

P. V. Thatayya Naidu wrote:
> 
> Hi,
>     Will  Object Serialistion and RMI work with this new release Kaffe 0.9.2 ?

I'm afraid not.  Any takers?

Cheers
Tim
--
  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
  London, UK.                           Email:   tim@tjwassoc.co.uk

From kaffe@kaffe.org  Tue Oct  7 03:22:13 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Tue, 07 Oct 1997 11:22:13 +0900
Subject: Working version of kaffe for BSDI 3.0?
In-Reply-To: Your message of Mon, 06 Oct 1997 09:25:54 MST.
Message-ID: <18855.876190933@rananim.ie.u-ryukyu.ac.jp>

Hi.

>What's the problem? I use BSD/OS 2.1 and running kaffe -version I get:

In BSD/OS 2.1 and 0.8.4, Kaffe works, but after 0.9.0, they use
stack offset in setjmp buf, which is not correnct in BSD/OS 3.0.
Ya, Bryan Felix was correct. It didn't work at all.

Here is a fix for 0.9.2

In config/i386/bsdi/md.h,  add SP_OFFSET redefine.

#include "i386/common.h"
#include "i386/threads.h"

#undef SP_OFFSET
#define SP_OFFSET               2

-- 
Shinji KONO @ Information Engineering, University of the Ryukyus
$B2OLn??<#(B @ $BN05eBg3X9)3XIt>pJs9)3X2J(B

From kaffe@kaffe.org  Tue Oct  7 07:45:29 1997
From: kaffe@kaffe.org (Orville R. Weyrich)
Date: Mon, 6 Oct 1997 23:45:29 -0700 (MST)
Subject: [FINALLY] Kaffe 0.9.2 released.
In-Reply-To: <3437A28F.41C67EA6@tjwassoc.co.uk>
Message-ID: <Pine.LNX.3.91.971006234218.28957C-100000@dopey.weyrich.com>

On Sun, 5 Oct 1997, Tim Wilkinson wrote:

>                 *** Kaffe is now JDK 1.1.2 compliant ***
 
> This version of Kaffe has the following new features:
> 
>  * Kaffe is now JDK 1.1.3 compliant !

Is one of these a typo, or is it compliant with both?

> What you need
> -------------

> 
> These two are are the minimum to get something useful working (you might
> already have the classes libraries from JDK 1.1.2; if so don't bother to
> get them again).

I have JDK 1.1.3 -- do I need to get 1.1.2, or will either work?

Thanks

orville

===================================================================
Orville R. Weyrich, Jr.                 Weyrich Computer Consulting
mailto:orville@weyrich.com                   http://www.weyrich.com
-------------------------------------------------------------------

From kaffe@kaffe.org  Tue Oct  7 13:48:38 1997
From: kaffe@kaffe.org (Gregor Hoffleit)
Date: Tue,  7 Oct 97 14:48:38 +0200
Subject: Kaffe 0.9.2 broken for NEXTSTEP
In-Reply-To: <3437A28F.41C67EA6@tjwassoc.co.uk>
References: <3437A28F.41C67EA6@tjwassoc.co.uk>
Message-ID: <9710071248.AA27266@mathi.uni-heidelberg.DE>

Kaffe 0.9.2 is broken for all NEXTSTEP architectures in various aspects, =
it won't compile (shared libs are inherently broken, trampoline code =
doesn't compile etc.pp.).


I thought I had sent Tim a NOTES/README.NEXTSTEP for inclusion in 0.9.2 =
that points out those problems and I sent in some patches, but somehow =
they didn't show up in 0.9.2.


Therefore I would opt to remove NEXTSTEP from the list of supported =
platforms for the time. Perhaps some future release can fix and =
re-install it.


For what it's worth, this is my page dedicated to kaffe on =
NEXTSTEP/OPENSTEP. You find working source and binary distributions of =
kaffe 0.8.4 and kaffe 0.9.0 for NEXTSTEP Intel, SPARC and m68k:


  http://www.mathi.uni-heidelberg.de/~flight/STEPkaffe


	Gregor



---

|
<smaller> Gregor Hoffleit                        Mathematisches =
Institut, Uni HD |

| flight@mathi.uni-heidelberg.de      INF 288, 69120 Heidelberg, Germany =
|

| (NeXTmail, MIME)                      (49)6221 54-5771    fax  54-8312 =
|

| PGP Key fingerprint =3D 23 8F B3 38 A3 39 A6 01  5B 99 91 D6 F2 AC CD =
C7 |

</smaller>

From kaffe@kaffe.org  Tue Oct  7 17:31:20 1997
From: kaffe@kaffe.org (Artur Biesiadowski)
Date: Tue, 7 Oct 1997 18:31:20 +0200 (MET DST)
Subject: Construction of primitive classes
Message-ID: <Pine.GSO.3.96L01AT.971007182047.20433A-100000@sunrise>

I'm trying to implement some of the missing native funs for
java.lang.reflect.
My question is about creating objects of one of the primitive classes -
Integer, Double, Boolean etc. 
I could created them by calling
execute_java_constructor( many args...)
but this would make quite a lot overhead - all this varargs and checks for
illegal access.
Second options is just to call  
newObject( classptr );
and the set value manually - but this could crash if class is not loaded??
Question: are primitive classes guaranteed to be loaded/compiled at this
time - calls for Field.get/Field.set are not supposed to be taken at
initialization I think.
If not, will it be enough just to check class->state and process it in
case it is not already ? 
Or just forget about speedup and stick to safe vararg constructors ?


Artur

From kaffe@kaffe.org  Tue Oct  7 17:44:08 1997
From: kaffe@kaffe.org (Vik Sohal)
Date: Tue, 07 Oct 1997 09:44:08 -0700
Subject: Object Seralistion!
Message-ID: <2.2.32.19971007164408.00c1df40@wizard.lynx.com>

Object Serialization requires an understanding of the low level protocol
that SUN uses in their own implementation. Up till a week ago this protocol
was not published, but I understand that it is now. We are looking at it and
are thinking about implementing it in Kaffe. If we do, I am quite certain we
will be able to post it as a patch to the Kaffe community.

Vik Sohal

At 11:34 PM 10/6/97 +0100, you wrote:
>P. V. Thatayya Naidu wrote:
>> 
>> Hi,
>>     Will  Object Serialistion and RMI work with this new release Kaffe
0.9.2 ?
>
>I'm afraid not.  Any takers?
>
>Cheers
>Tim
>--
>  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
>  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
>  London, UK.                           Email:   tim@tjwassoc.co.uk
>
>
>
***************************************************************
* Vik Sohal                     *  email: vik@lynx.com        *
* Lynx Real-Time Systems        *  voice: (408) 879-3900 x119 *
* 2239 Samaritan Drive          *    fax: (408) 879-3920      *
* San Jose, CA 95124            *                             *
* ------------------------------*-----------------------------*
* cute msg: "We just turned it on, and it's I.Q. is already   *
*             in the thousands!"                              *
***************************************************************

From kaffe@kaffe.org  Tue Oct  7 18:26:57 1997
From: kaffe@kaffe.org (Per Bothner)
Date: Tue, 07 Oct 1997 10:26:57 -0700
Subject: Object Seralistion!
In-Reply-To: Your message of "Tue, 07 Oct 1997 09:44:08 PDT."
 <2.2.32.19971007164408.00c1df40@wizard.lynx.com>
Message-ID: <199710071726.KAA00461@cygnus.com>

> Up till a week ago this protocol was not published,

It has been published for a while (with the Serialization "white paper"),
(though I did not realize it until someone pointed it out).

> but I understand that it is now.

Perhaps there is an improved description.

Thre are rumors that Sun is changing RMI to be more compatible with CORBA.
I don't know what this entails.  Even if they do, the old protocol will
still be useful for quite a while.

	--Per Bothner
Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner

From kaffe@kaffe.org  Tue Oct  7 23:27:30 1997
From: kaffe@kaffe.org (Alexandre Oliva)
Date: 07 Oct 1997 20:27:30 -0200
Subject: Installing kaffe on IRIX 5.2 with GNU as
Message-ID: <orafgl43al.fsf@sunsite.dcc.unicamp.br>

Hi there!

I had trouble for compiling kaffe 0.9.2 on IRIX 5.2 because GNU as was
used instead of /usr/bin/as.  The object-file threads.o produced by
GNU as wouldn't link, whereas the one produced by /usr/bin/as would.
I'd suggest that, for the particular case of IRIX, the Makefile
fragment contained a definition such as:

AS=/usr/bin/as

-- 
Alexandre Oliva
mailto:oliva@dcc.unicamp.br mailto:aoliva@acm.org
http://www.dcc.unicamp.br/~oliva
Universidade Estadual de Campinas, SP, Brasil
From kaffe@kaffe.org  Tue Oct  7 22:10:12 1997
From: kaffe@kaffe.org (Tim Wilkinson)
Date: Tue, 07 Oct 1997 22:10:12 +0100
Subject: Object Seralistion!
References: <2.2.32.19971007164408.00c1df40@wizard.lynx.com>
Message-ID: <343AA534.41C67EA6@tjwassoc.co.uk>

Vik Sohal wrote:
> 
> Object Serialization requires an understanding of the low level protocol
> that SUN uses in their own implementation. Up till a week ago this protocol
> was not published, but I understand that it is now. We are looking at it and
> are thinking about implementing it in Kaffe. If we do, I am quite certain we
> will be able to post it as a patch to the Kaffe community.

I've actually looked into this and found that the documentation for the
original JDK 1.1-beta contains more than enough information on how to
write Serialization.  You should bear in mind that most of the hard work
will be done in the Java code and only the native stubs need to be
written - and these don't worry about the format of Serialized data
anyway.  Of course, the actual function of these native methods isn't
documented so you have to guess ...  But then, thats how the rest of the
native stuff was written.

Cheers
Tim
--
  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
  London, UK.                           Email:   tim@tjwassoc.co.uk

From kaffe@kaffe.org  Tue Oct  7 11:04:03 1997
From: kaffe@kaffe.org (Ralf W. Stephan)
Date: Tue, 7 Oct 1997 11:04:03 +0100
Subject: Object Seralistion!
In-Reply-To: <3439677C.3F54BC7E@tjwassoc.co.uk>; from Tim Wilkinson on Mon, Oct 06, 1997 at 11:34:36PM +0100
References: <199710061721.KAA01323@faith> <3439677C.3F54BC7E@tjwassoc.co.uk>
Message-ID: <19971007110403.51428@ark.franken.de>

Tim wrote 
> P. V. Thatayya Naidu wrote:
> >     Will  Object Serialistion and RMI work with this new release Kaffe 0.9.2 ?
> I'm afraid not.  Any takers?

No, I won't bite yet, I'm still trying to understand the source.
Maybe someone can help me:

- how do you know which native functions are needed by Java's classes?
  do you just wait for LinkErrors or is there something like an API?
- many files in packages/tjwassoc... are machine generated --- from
  what input?
- can the JITC in principle take a bytecode call to a member func
  and instead direct it to a native implementaton of the class?
  Example: java.util.Vector: could this be speeded up by *directly*
  calling my own C implementation (possibly without bounds checking, e.g.)?
- is there a point in rewriting some classes to get better performance /
  free versions of them, or is this outside the kaffe project?

Thanks,
ralf
-- 
URL: http://rws.home.pages.de/

From kaffe@kaffe.org  Wed Oct  8 20:51:45 1997
From: kaffe@kaffe.org (Darrell Gallion)
Date: Wed, 08 Oct 1997 15:51:45 -0400
Subject: Configure for QNX ??
Message-ID: <343BE451.B0D6847F@cldx.com>

After running configure, make fails.
Memory model and size of int, long, .. are wrong. And as far as I know
the Watcom compiler has no 64 bit
longs.

Thanx in advance for any insight.

Darrell Gallion
dgallion@cldx.com


From kaffe@kaffe.org  Thu Oct  9 00:02:56 1997
From: kaffe@kaffe.org (Stewart Allen)
Date: Wed, 8 Oct 1997 19:02:56 -0400 (EDT)
Subject: cannot build 0.92 on linux 2.0.30
Message-ID: <Pine.LNX.3.95.971008185534.24163A-100000@flex.neuron.com>


 About 3 seconds into the build I get:


make[4]: Entering directory
`/local/hd1/home/stewart/build/kaffe-0.9.2/packages/tjwassoc.co.uk/APIcore/lib'
gcc -g -O2 -I. -I../../../../include -I./../../../../include
-I../../../../config -I./../../../../config -c -fPIC ./java.lang/Math.c
In file included from ./java.lang/Math.c:16:
../../../../config/config-math.h:24: #error "Need some form of remainder"
make[4]: *** [Math.o] Error 1
make[4]: Leaving directory
`/local/hd1/home/stewart/build/kaffe-0.9.2/packages/tjwassoc.co.uk/APIcore/lib'


 Which I traced down to the fact that Configure incorrectly determines
 that the system doesn't support "fmod" or "drem" (both of which are
 present). I can correct these settings in config.h at which point the
 entire thing builds until the kaffevm link which gives:


gcc -g -O2  -I. -I./../kaffevm -I../../config -I./../../config
-I../../include -I./../../include -o kaffe main.o -L../kaffevm -lkaffevm
../kaffevm/libkaffevm.so: undefined reference to `remainder'
make[2]: *** [kaffe] Error 1


 Any suggestions?

 -stewart-

From kaffe@kaffe.org  Thu Oct  9 02:56:24 1997
From: kaffe@kaffe.org (Glynn Clements)
Date: Thu, 9 Oct 1997 02:56:24 +0100
Subject: cannot build 0.92 on linux 2.0.30
In-Reply-To: <Pine.LNX.3.95.971008185534.24163A-100000@flex.neuron.com>
References: <Pine.LNX.3.95.971008185534.24163A-100000@flex.neuron.com>
Message-ID: <199710090156.CAA08267@cerise.sensei.co.uk>


Stewart Allen wrote:

>  About 3 seconds into the build I get:
> 
> 
> make[4]: Entering directory
> `/local/hd1/home/stewart/build/kaffe-0.9.2/packages/tjwassoc.co.uk/APIcore/lib'
> gcc -g -O2 -I. -I../../../../include -I./../../../../include
> -I../../../../config -I./../../../../config -c -fPIC ./java.lang/Math.c
> In file included from ./java.lang/Math.c:16:
> .../../../../config/config-math.h:24: #error "Need some form of remainder"
> make[4]: *** [Math.o] Error 1
> make[4]: Leaving directory
> `/local/hd1/home/stewart/build/kaffe-0.9.2/packages/tjwassoc.co.uk/APIcore/lib'
> 
> 
>  Which I traced down to the fact that Configure incorrectly determines
>  that the system doesn't support "fmod" or "drem" (both of which are
>  present). I can correct these settings in config.h at which point the
>  entire thing builds until the kaffevm link which gives:
> 
> 
> gcc -g -O2  -I. -I./../kaffevm -I../../config -I./../../config
> -I../../include -I./../../include -o kaffe main.o -L../kaffevm -lkaffevm
> .../kaffevm/libkaffevm.so: undefined reference to `remainder'
> make[2]: *** [kaffe] Error 1

This indicates that remainder isn't being replaced with fmod.

Make sure that HAVE_REMAINDER is undefined and HAVE_FMOD is defined,
e.g. insert

	#undef HAVE_REMAINDER
	#define HAVE_FMOD 1

at the top of config-math.h.

I'm running Linux 2.0.30 with libc 5.4.23, and it compiled out of the
box for me.

-- 
Glynn Clements <glynn@sensei.co.uk>

From kaffe@kaffe.org  Thu Oct  9 02:56:24 1997
From: kaffe@kaffe.org (Glynn Clements)
Date: Thu, 9 Oct 1997 02:56:24 +0100
Subject: cannot build 0.92 on linux 2.0.30
In-Reply-To: <Pine.LNX.3.95.971008185534.24163A-100000@flex.neuron.com>
References: <Pine.LNX.3.95.971008185534.24163A-100000@flex.neuron.com>
Message-ID: <199710090156.CAA08267@cerise.sensei.co.uk>


Stewart Allen wrote:

>  About 3 seconds into the build I get:
> 
> 
> make[4]: Entering directory
> `/local/hd1/home/stewart/build/kaffe-0.9.2/packages/tjwassoc.co.uk/APIcore/lib'
> gcc -g -O2 -I. -I../../../../include -I./../../../../include
> -I../../../../config -I./../../../../config -c -fPIC ./java.lang/Math.c
> In file included from ./java.lang/Math.c:16:
> .../../../../config/config-math.h:24: #error "Need some form of remainder"
> make[4]: *** [Math.o] Error 1
> make[4]: Leaving directory
> `/local/hd1/home/stewart/build/kaffe-0.9.2/packages/tjwassoc.co.uk/APIcore/lib'
> 
> 
>  Which I traced down to the fact that Configure incorrectly determines
>  that the system doesn't support "fmod" or "drem" (both of which are
>  present). I can correct these settings in config.h at which point the
>  entire thing builds until the kaffevm link which gives:
> 
> 
> gcc -g -O2  -I. -I./../kaffevm -I../../config -I./../../config
> -I../../include -I./../../include -o kaffe main.o -L../kaffevm -lkaffevm
> .../kaffevm/libkaffevm.so: undefined reference to `remainder'
> make[2]: *** [kaffe] Error 1

This indicates that remainder isn't being replaced with fmod.

Make sure that HAVE_REMAINDER is undefined and HAVE_FMOD is defined,
e.g. insert

	#undef HAVE_REMAINDER
	#define HAVE_FMOD 1

at the top of config-math.h.

I'm running Linux 2.0.30 with libc 5.4.23, and it compiled out of the
box for me.

-- 
Glynn Clements <glynn@sensei.co.uk>
From kaffe@kaffe.org  Thu Oct  9 10:10:55 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Thu, 9 Oct 1997 10:10:55 +0100
Subject: Configure for QNX ??
In-Reply-To: <343BE451.B0D6847F@cldx.com> (message from Darrell Gallion on
 Wed, 8 Oct 1997 20:51:45 +0100)
Message-ID: <9710091010.aa11291@stlind.lint.lyngso-industri.dk>


> After running configure, make fails.
> Memory model and size of int, long, .. are wrong. And as far as I know
> the Watcom compiler has no 64 bit
> longs.

Watcom 11 has 64 bit ints BUT you must use gcc for compiling kaffe on
QNX. 

Note that kaffe-0.9.2 does not compile out of the box for QNX. 

0.9.1, however, does with a little help. 

Here is the message that I posted back in july to comp.os.qnx,
describing how to compile kaffe-0.9.1 on QNX.

---------------------------------------------------------------------

I have ported the kaffe virtual machine to QNX-4.23. 

Kaffe is a free implementation of the Java virtual machine, written by
Tim Wilkinson. Kaffe uses Just-In-Time compilation of byte-code into
machine code which makes it pretty efficient (considerably faster than
Sun's "reference" interpreter based implementation - slightly slower
than Sun's JIT based implementation).

Kaffe can be used together with Sun's class libraries version
1.1.2. The complete API is not yet supported by native methods though
- I think more or less the subset corresponding to version 1.0.2 is
supported. An important exception is that awt (i.e., graphics) is not
supported by kaffe-0.9.1. 

Gnu CC 2.7.2 is required in order to compile kaffe. gcc can be
obtained from http://www.teaser.fr/~jcmichot/gcc.

The QNX patches have already been incorporated into kaffe-0.9.1, which
can be obtained from ftp://ftp.kaffe.org/pub/kaffe/kaffe-0.9.1.tgz. 

A single bugfix is required to make it compile: in
config/i386/trampolines.c replace ".global" with ".globl"

Sun's class libraries are available from a number of places, either as
part of kaffe (kaffe-0.9.1-package-javasoft.tgz), or as part of
jdk-1.1.2 which can be downloaded from javasoft.com.


Regards, Jeppe Sommer
jso@lyngso-industri.dk

----------------------------------------------------------------------

One more note: the ld command provided with gcc will confuse kaffe's
configure script because it exits zero even when unresolved symbols
are encountered. A fixed (beta) ld can be obtained from 
http://www.teaser.fr/~jcmichot/gcc/ld. I suspect that this will be a
temporary issue. 



Jeppe

From kaffe@kaffe.org  Thu Oct  9 21:42:19 1997
From: kaffe@kaffe.org (Alain Magloire)
Date: Thu, 9 Oct 1997 16:42:19 -0400 (EDT)
Subject: {K/J}NI
In-Reply-To: <3438E1BB.1CFBAE39@tjwassoc.co.uk> from "Tim Wilkinson" at Oct 6, 97 02:03:55 pm
Message-ID: <199710092042.QAA01026@pct62.cae.ca>

Bonjour M. Wilkinson

  Is kaffe supporting JNI, if not is there plan for it ?

I see that kaffe uses the old stub interface 

Thanks


--
alain

From kaffe@kaffe.org  Fri Oct 10 20:59:05 1997
From: kaffe@kaffe.org (Amitabh Arora)
Date: Fri, 10 Oct 1997 12:59:05 -0700
Subject: kaffe port for HP-UX 9.05 OS
Message-ID: <01BCD57C.4AF62E80@amitabh@aspec.com>

Has anyone done this port?  I do not need the JIT feature.
So the port should be easy.

Thanks. -amitabh arora
From kaffe@kaffe.org  Sat Oct 11 00:34:46 1997
From: kaffe@kaffe.org (James A. Hillyerd)
Date: Fri, 10 Oct 1997 16:34:46 -0700 (PDT)
Subject: JavaLobby and JavaOS
Message-ID: <Pine.LNX.3.95.971010163139.30292A-100000@animal.blarg.net>

There is some interest on the JavaLobby to write a JavaOS.  This would
probably be based on the Mach microkernel, and perhaps some pieces of GNU
Hurd.  The thought was that Kaffe would be embedded into this.  The
working name seems to be 'Oz': as in the land where anything can happen.

We could really use some C hackers, and some Kernel hackers.  Please join
in the discussion, and you should also consider joining the Java Lobby,
it's free!

http://www.javalobby.com/

Sorry if this irritates anyone.  Nobody asked me to post this here, so
direct all flames to me. =)

-james

[  James A. Hillyerd (JH2162) - james@blarg.net - Web Developer  ]
[  http://www.blarg.net/~james/   http://www.hyperglyphics.com/  ]
[ 1024/B11C3751 CA 1C B3 A9 07 2F 57 C9  91 F4 73 F2 19 A4 C5 88 ]

From kaffe@kaffe.org  Sat Oct 11 12:42:32 1997
From: kaffe@kaffe.org (Fletcher E Kittredge)
Date: Sat, 11 Oct 1997 07:42:32 -0400
Subject: JavaLobby and JavaOS
In-Reply-To: Your message of "Fri, 10 Oct 1997 16:34:46 PDT."
 <Pine.LNX.3.95.971010163139.30292A-100000@animal.blarg.net>
Message-ID: <199710111142.HAA14774@river.gwi.net>

On Fri, 10 Oct 1997 16:34:46 -0700 (PDT)  "James A. Hillyerd" wrote:
> There is some interest on the JavaLobby to write a JavaOS.  This would
> probably be based on the Mach microkernel, and perhaps some pieces of GNU
> Hurd.  The thought was that Kaffe would be embedded into this.  The
> working name seems to be 'Oz': as in the land where anything can happen.
> 
> We could really use some C hackers, and some Kernel hackers.  Please join
> in the discussion, and you should also consider joining the Java Lobby,
> it's free!

As an ex-Mach and C hacker, let me say that there are real good
technical reasons that Mach and Hurd failed.  The micro-kernel
architecture is dog slow; adding Java to the mix is not going to speed
it up.

Before wasting good Java developer time on this, please have someone
you trust go back and read all the papers from the late 80's, early
90's which detail the significant performance problems with a
micro-kernel architecture...

regards,
fletcher

From kaffe@kaffe.org  Sat Oct 11 14:26:05 1997
From: kaffe@kaffe.org (Vijay Saraswat)
Date: Sat, 11 Oct 1997 09:26:05 -0400 (EDT)
Subject: JavaLobby and JavaOS
In-Reply-To: "James A. Hillyerd"'s message of Fri, 10 Oct 1997 16:34:46 -0700 (PDT)
Message-ID: <199710111326.JAA23400@radish.research.att.com>

Oz is the name of a programming system produced by the German AI
institute (DFKI). 

From kaffe@kaffe.org  Sat Oct 11 15:51:01 1997
From: kaffe@kaffe.org (Sean McDirmid)
Date: Sat, 11 Oct 1997 07:51:01 -0700 (PDT)
Subject: JavaLobby and JavaOS
In-Reply-To: <199710111142.HAA14774@river.gwi.net>
Message-ID: <Pine.OSF.3.96.971011074458.4584A-100000@sanjuan.cs.washington.edu>

On Sat, 11 Oct 1997, Fletcher E  Kittredge wrote:

> As an ex-Mach and C hacker, let me say that there are real good
> technical reasons that Mach and Hurd failed.  The micro-kernel
> architecture is dog slow; adding Java to the mix is not going to speed
> it up.
> 
> Before wasting good Java developer time on this, please have someone
> you trust go back and read all the papers from the late 80's, early
> 90's which detail the significant performance problems with a
> micro-kernel architecture...

Now hold on here, if I understand it, the micro-kernal architecture was
slow b/c of the way it kept its form of modularity, through runtime
barriers and abstractions.  The great thing about Java code is that it is
safe and this safety can be shown (lacking in kaffe but I've written a
verifier for kimera - kimera.cs.washington.edu).  Several good performing
(and commercial) operating systems have been "derived" from
micro-kernal's, such as NT, NextStep and perhaps the upcoming BeOS. How
about a derived micro-kernal that takes Java code, compiles it, and places
it close to the micro-kernal as a trusted extension?  Now we don't have a
micro-kernal anymore but you might be able to get the performance you are
looking for. You still get the safety and abstraction that the original
micro-kernal provided (through the virtual machine compiled to machine
code interface).

Sean McDirmid
kimera.cs.washington.edu


From kaffe@kaffe.org  Sat Oct 11 17:28:18 1997
From: kaffe@kaffe.org (Bruce R. Montague)
Date: Sat, 11 Oct 1997 09:28:18 -0700
Subject: JavaLobby and JavaOS
Message-ID: <199710111628.JAA02908@madrone.cse.ucsc.edu>

 re microkernel OS engineering/performance problems - 2 recent papers 
 that may prove of interest in this regard are:

 "Experience with the Development of a Microkernel-Based, Multiserver
  Operating System", Freeman Rawson, Proc. 6th Workshop on Hot Topics
  in Operating Systems, 5-6 May 1997 (HOTOS);
 
 "The Failure of Personalities to Generalize", Brett Fleisch, ditto...

 These are also intrinsically of interest as IBM apparently lost
 $2 _billion_ on this... Quote: "Despite the level of investment and the
 adoption of what was deemed to be the best of operating systems research
 and advanced development of the 1980s, the project, in retrospect, was
 not a technical success." (Rawson).

 Apologies if this is off subject, I'm just seconding the motion to 
 think this through carefully; forewarned is forarmed, ... investing
 the effort in things like porting interfaces, validation/compatibility
 tests, etc., might be more effective?

 Regards,
 bruce

From kaffe@kaffe.org  Sat Oct 11 17:55:45 1997
From: kaffe@kaffe.org (Markus Peter)
Date: Sat, 11 Oct 1997 18:55:45 +0200
Subject: JavaLobby and JavaOS
Message-ID: <01bcd666$849cd450$0201a8c0@warpnt.spin.de>

Ok. I guess it's also time for a statement from you.
I was one of the initiators of this project.

In the mean time, we already have found somebody,
who will create a new kernel specifically for a Java OS.
This one will be very small, as a Java based OS eliminates
a lot of needs (process switching, protected memory etc...)

As soon as we have that OS, we will need a port of Kaffe
for it - I've chosen Kaffe as JIT Compiler for this OS,
because it's free and because I once needed a JIT Compiler
for Linux, and I was very disappointed of the Blackdown port,
while Kaffe appeared very excellent to me (even at that time
- it was version 0.8.4).

Unfortunately the person who will create our OS cannot
port Kaffe himself, as he's got a Java source code license
from Sun, and we want a Java OS free from any copyrights
ASAP.

That's where we will need some help from one of you for sure.
Most of the volunteers right now are experienced Java programmers,
but they do not have enough C experience to port Kaffe or it
is too long ago they used C. So if one of you is interested in
this project and would be willing to port kaffe to that OS kernel,
as soon as it is created, please contact me.

I guess without the help of one of you this project is determined
to die.

(Hehe - please do not flame me with telling me, that I should
not try to create an OS without having a clue of it, in the first place :)

Markus Peter (Developer), SPiN GmbH
warpi@spin.de   -    http://www.spin.de
Java, Perl and Web Site Development
Unite for Java!  - http://www.javalobby.org/


From kaffe@kaffe.org  Sat Oct 11 18:25:05 1997
From: kaffe@kaffe.org (roy)
Date: Sat, 11 Oct 1997 10:25:05 -0700
Subject: JavaLobby and JavaOS
In-Reply-To: <Pine.OSF.3.96.971011074458.4584A-100000@sanjuan.cs.washing
 ton.edu>
References: <199710111142.HAA14774@river.gwi.net>
Message-ID: <3.0.3.32.19971011102505.009652e0@mail.rdc1.sfba.home.net>


A microkernel is probably the worst possible fit for a 
"pure Java OS" (something that supports only Java)
if you are going to maintain the microkernel view of
things and run Kaffe as a server or subsystem.

Due to Java's security and safety, a conventional OS
carries a lot of extra baggage that is unnecessary for
a "pure Java OS".
 
The objective of microkernels is to isolate and
optimize mechanism in a small core, while providing
for multiple policies in user space. But with a 
"pure Java OS" this is unnecessary, since all you
need to support is Java. Furthermore, with a verifier
to assist, you may not need a user/kernel distinction
any more.

Instead what one probably should do is link Kaffe directly
into the kernel of some free OS (preferably real-time),
then throw out all the unnecessary stuff (user-kernel
distinction, processes, etc...)

A number of commercial and research OS vendors have used
this approach (including Chorus and the Univ. of Utah. Except
that neither is real-time, which may be important for us.)
i.e. to turning a conventional OS into a "pure java OS"
involves mostly paring stuff away.

Come to think of it, a micro-kernel may be a good starting
point if we were to link Kaffe directly into it (instead of
making it a "single server" or a "subsystem"), then turn
the non-Kaffe portions into a "pico-kernel" :-)

It may be that single-address space OSs are the best
match for Java. Perhaps we should look at Opal. Or a
kernel-extensible OS like Spin. These would probably
a much better starting point than Mach or Hurd.

By the way, I think Next is based on Mach 2.5, which is
not a microkernel. Neither is the current release of NT.

Roy


From kaffe@kaffe.org  Sat Oct 11 19:18:15 1997
From: kaffe@kaffe.org (Markus Peter)
Date: Sat, 11 Oct 1997 20:18:15 +0200
Subject: JavaOS
Message-ID: <01bcd672$0abc9c40$0201a8c0@warpnt.spin.de>

I tried emailing this mailing list earlier, but strangely, my message
did not arrive - well - here it is again - I hope you now do not
get two copies of it.

Ok. I guess it's also time for a statement from me.
I was one of the initiators of this project.

In the mean time, we already have found somebody,
who will create a new kernel specifically for a Java OS.
This one will be very small, as a Java based OS eliminates
a lot of needs (process switching, protected memory etc...)

As soon as we have that OS, we will need a port of Kaffe
for it - I've chosen Kaffe as JIT Compiler for this OS,
because it's free and because I once needed a JIT Compiler
for Linux, and I was very disappointed of the Blackdown port,
while Kaffe appeared very excellent to me (even at that time
- it was version 0.8.4).

Unfortunately the person who will create our OS cannot
port Kaffe himself, as he's got a Java source code license
from Sun, and we want a Java OS free from any copyrights
ASAP.

That's where we will need some help from one of you for sure.
Most of the volunteers right now are experienced Java programmers,
but they do not have enough C experience to port Kaffe or it
is too long ago they used C. So if one of you is interested in
this project and would be willing to port kaffe to that OS kernel,
as soon as it is created, please contact me.
I guess without the help of one of you this project is determined
to die.

(Hehe - please do not flame me with telling me, that I should
not try to create an OS without having a clue of it, in the first place :)

And one last note to Bruce Montague:
Hurd did not fail yet - in fact the first versions of Hurd were released
some months ago.
And yes - I know, lots of people say, microkernel designs have failed,
but on the other hand there are also papers citing that they are useful.
WinNT uses the MachKernel, NextStep does, Rhapsody will do.
But that does not matter anymore - after all we now have found
a developer who will create an exclusive kernel for us.

I wanted to use a microkernel in the beginning, because most
people participating in the project at the beginning were only Java
developers but were not very experienced with C programming.
So I though we should take something existing, maybe a micro-kernel,
because it is not bloated too much with unnecessary stuff.
After all a true kernel for a Java OS need not do much.


Markus Peter (Developer), SPiN GmbH
warpi@spin.de   -    http://www.spin.de
Java, Perl and Web Site Development
Unite for Java!  - http://www.javalobby.org/

From kaffe@kaffe.org  Sun Oct 12 09:37:25 1997
From: kaffe@kaffe.org (Gabriel Sebestyen)
Date: Sun, 12 Oct 1997 10:37:25 +0200
Subject: JavaOS
References: <01bcd672$0abc9c40$0201a8c0@warpnt.spin.de>
Message-ID: <34408C45.3041166C@accentcomm.com>

Hi,

First of all, Windows NT(tm :-) ) uses an own kernel developed by DEC
not Mach. Mach is designed to be a small and independent microkernel
which can be used to build your own system on. Mach 3 finally lost the
only (BSD) UN*X dependent part now it has a BSD server instead.
Some facts about Mach:
* unified and VERY efficient message handling
* simple but efficient virtual memory handling (who likes NT's memory
management? :-) )
* supports threads
* ported to a lot of platforms but it's as independent from them as it
can be

I think Mach is a real microkernel (because it was designed to be
originally and choosen by a lot of OS's) and an excellent starting point
to develope e.g. a JavaOS. Making a new kernel from the beginning does
not make sense! Just a little performance example:
An HP UNIX emulator to Mach wrote by a team for test purposes was faster
than the original HP! It's true believe me!

Gabriel Sebestyen


From kaffe@kaffe.org  Mon Oct 13 07:57:34 1997
From: kaffe@kaffe.org (Luis L. Fernandez)
Date: Mon, 13 Oct 1997 08:57:34 +0200
Subject: JavaOS
Message-ID: <199710130657.IAA00386@diogenes.iit.upco.es>


Hi,

	
	I've listening your discussion about a new Java-OS and
micro-kernels design....

	Well, in the Computer Science Department at Utah University
you can find the answer to this discussion. They have already
developed a Java-OS using a library with provides you a toolkit to
build OS named OSkit upon and Intel ix86 architecture. I've used it
and it's really good. The OSKit code it's free and it's available in

http://www.cs.utah.edu/projects/flux/oskit/

They are kernel hackers and they have redesigned the Mach micro-kernel
to run fine (search about Fluke project).

I hope this message seems useful to somebody ;-)

Best Regards,

	Luis.

---------
Luis Lopez Fernandez
e-mail: llf@computer.org

From kaffe@kaffe.org  Mon Oct 13 12:17:31 1997
From: kaffe@kaffe.org (Vijay Saraswat)
Date: Mon, 13 Oct 1997 07:17:31 -0400 (EDT)
Subject: JavaLobby and JavaOS
In-Reply-To: Sean McDirmid's message of Sat, 11 Oct 1997 07:51:01 -0700 (PDT)
Message-ID: <199710131117.HAA03790@radish.research.att.com>

   X-Authentication-Warning: vega.soi.city.ac.uk: lp set sender to owner-kaffe@soi.city.ac.uk using -f
   From: Sean McDirmid <mcdirmid@cs.washington.edu>
   Sender: owner-kaffe@soi.city.ac.uk

   On Sat, 11 Oct 1997, Fletcher E  Kittredge wrote:

   > As an ex-Mach and C hacker, let me say that there are real good
   > technical reasons that Mach and Hurd failed.  The micro-kernel
   > architecture is dog slow; adding Java to the mix is not going to speed
   > it up.
   > 
   > Before wasting good Java developer time on this, please have someone
   > you trust go back and read all the papers from the late 80's, early
   > 90's which detail the significant performance problems with a
   > micro-kernel architecture...

   Now hold on here, if I understand it, the micro-kernal architecture was
   slow b/c of the way it kept its form of modularity, through runtime
   barriers and abstractions.  The great thing about Java code is that it is
   safe and this safety can be shown (lacking in kaffe but I've written a
   verifier for kimera - kimera.cs.washington.edu).  Several good performing
   (and commercial) operating systems have been "derived" from
   micro-kernal's, such as NT, NextStep and perhaps the upcoming BeOS. How
   about a derived micro-kernal that takes Java code, compiles it, and places
   it close to the micro-kernal as a trusted extension?  Now we don't have a
   micro-kernal anymore but you might be able to get the performance you are
   looking for. You still get the safety and abstraction that the original
   micro-kernal provided (through the virtual machine compiled to machine
   code interface).

I agree with Sean here. The notion of a strongly-typed,
dynamically linked language, like Java, makes some fresh
approaches possible. 

To go a step beyond what Sean is saying: in fact it is possible
(the "declarative bytecode verification" approach) to develop
bytecode verifiers *systematically*, using a first-principles
approach, from a constraint-based specification of the opcodes
involved. One can translate an input byte code program, P,
compositionally (byte-code for byte-code) into a constraint
program, T(P) which "obviously" states the constraints on the
run-time type-state of the Java Virtual Machine, which must hold
for program execution to be type-safe. T(P) is "loop-free" and
will take time proportional to its length, which is proportional
to the length of P, to execute. If T(P) does not deadlock or
produce "false", P is type-safe.

This approach differs slightly from the one that Sean and
Javasoft have implemented. In their cases, there are separate
*byte code verifiers*, usually largish (10-15 page) C programs,
implementing a "data-flow" analysis, documented by 10-15 pages of
informal English (I have not seen Sean's documentation yet),
about whose correctness it is hard to say anything formal. 

In any case, the bottom line for this list should be, I think,
that one can assume that the subproblem of verifying the
type-safety of JVM-like bytecode is solved, with of course, one
important caveat: native code. (A lot of the "core" classes in
Java in fact contain natively implemented methods, and one has to
use separate techniques for verifying them.) 

I would not be too surprised if this approach (Java's approach)
of doing "link-time" (i.e. first-instruction-execution-time)
type-checking actually resulted in significant performance
*improvements* over standard OS's. One thing to keep in mind is
that Java's architecture is such that the bytecode verifier can
in fact accept some input code *even if it is not type-safe in
isolation* --- as long as it is type-safe when combined with the
other classes that are already loaded. That is, there are some
possibilities of doing inter-classfile optimizations (e.g.
eliminating some classcasts generated by the compiler) in a
typesafe way, that can produce even more performance benefits
than are posible if one does "straight" Java verification, as
done currently by Sun's Java systems.

Best,
Vijay

From kaffe@kaffe.org  Mon Oct 13 12:10:10 1997
From: kaffe@kaffe.org (Roger Binns)
Date: Mon, 13 Oct 1997 11:10:10 +0000 (GMT)
Subject: JavaLobby and JavaOS
In-Reply-To: <01bcd666$849cd450$0201a8c0@warpnt.spin.de> from "Markus Peter" at Oct 11, 97 06:55:45 pm
Message-ID: <199710131110.LAA25110@avon.x.co.uk>

> In the mean time, we already have found somebody,
> who will create a new kernel specifically for a Java OS.
> This one will be very small, as a Java based OS eliminates
> a lot of needs (process switching, protected memory etc...)

One thing to be careful of is drivers.  I don't know if you plan
to support commodity PC hardware, or will have very few supported
devices.

In any case, my suggestion would be to take a freeware UNIX (eg Linux)
that has lots of device support, and slowly remove/rewrite bits
to get to your goal.  This has the advantage that you have a working system 
from day one.  

(Ironically, it would also mirror the way Linux was built from Minix!)

Roger
-- 
Roger Binns   rogerb@sco.com | The reasonable man adapts himself to the world;
Software Architect           | the unreasonable one persists in trying to
Client Integration Division  | to adapt the world to himself. Therefore all
SCO, Vision Park, Cambridge  | progress depends on the unreasonable man - GBS

From kaffe@kaffe.org  Mon Oct 13 15:12:20 1997
From: kaffe@kaffe.org (Aleksey Sudakov)
Date: Mon, 13 Oct 97 10:12:20 -0400
Subject: JavaOS
In-Reply-To: <199710130657.IAA00386@diogenes.iit.upco.es>
References: <199710130657.IAA00386@diogenes.iit.upco.es>
Message-ID: <9710131412.AA07422@rkinc.com>

: 	Well, in the Computer Science Department at Utah University
: you can find the answer to this discussion. They have already
: developed a Java-OS using a library with provides you a toolkit to
: build OS named OSkit upon and Intel ix86 architecture. I've used it
: and it's really good. The OSKit code it's free and it's available in

It's totally amazing. The only bad things are that it's not available for  
immediate download and it's not free. They say that it's free for  
non-commercial use and IMHO it will affect acceptance of this totally cool  
idea.

I wonder when free software become really free.

Regards,
Aleksey

From kaffe@kaffe.org  Mon Oct 13 15:29:50 1997
From: kaffe@kaffe.org (Vijay Saraswat)
Date: Mon, 13 Oct 1997 10:29:50 -0400 (EDT)
Subject: JavaLobby and JavaOS
In-Reply-To: "Bruce R. Montague"'s message of Sat, 11 Oct 1997 09:28:18 -0700
Message-ID: <199710131429.KAA05876@radish.research.att.com>

RE: the HOTOS talks, a URL is

  http://www.eecs.harvard.edu/hotos/scribes/sessions/implementation-1.html


   Date: Sat, 11 Oct 1997 09:28:18 -0700
   From: "Bruce R. Montague" <brucem@cse.ucsc.edu>

    re microkernel OS engineering/performance problems - 2 recent papers 
    that may prove of interest in this regard are:

    "Experience with the Development of a Microkernel-Based, Multiserver
     Operating System", Freeman Rawson, Proc. 6th Workshop on Hot Topics
     in Operating Systems, 5-6 May 1997 (HOTOS);

    "The Failure of Personalities to Generalize", Brett Fleisch, ditto...

From kaffe@kaffe.org  Mon Oct 13 16:01:38 1997
From: kaffe@kaffe.org (Thanh Ma)
Date: Mon, 13 Oct 1997 11:01:38 -0400
Subject: JavaOS
In-Reply-To: <9710131412.AA07422@rkinc.com>; from Aleksey Sudakov on Mon, Oct 13, 1997 at 10:12:20AM -0400
References: <199710130657.IAA00386@diogenes.iit.upco.es> <9710131412.AA07422@rkinc.com>
Message-ID: <19971013110137.07019@dino.encore.com>

On Mon, Oct 13, 1997 at 10:12:20AM -0400, Aleksey Sudakov wrote:
> 
> It's totally amazing. The only bad things are that it's not available for  
> immediate download and it's not free. They say that it's free for  
> non-commercial use and IMHO it will affect acceptance of this totally cool  
> idea.
> 
> I wonder when free software become really free.

SCO's commercial UnixWare and OpenServer are 'free' for non-commercial
purposes :-)

Thanh

From kaffe@kaffe.org  Mon Oct 13 16:38:00 1997
From: kaffe@kaffe.org (Patrick Logan)
Date: Mon, 13 Oct 97 08:38 PDT
Subject: JavaLobby and JavaOS
In-Reply-To: <Pine.LNX.3.95.971010163139.30292A-100000@animal.blarg.net>
References: <Pine.LNX.3.95.971010163139.30292A-100000@animal.blarg.net>
Message-ID: <m0xKmZG-000DM5C@gharlane.gemstone.com>

>>>>> "James" == James A Hillyerd <james@blarg.net> writes:

    James> There is some interest on the JavaLobby to write a JavaOS.
    James> This would probably be based on the Mach microkernel, and
    James> perhaps some pieces of GNU Hurd.

Someone from Utah already ported kaffe to their OS framework. This
framework was developed by former Mach developers to *replace* Mach
because it was too unwieldy. Check it out in the research pages at
www.cs.utah.edu, I believe.

-- 
Patrick Logan                 mailto:patrickl@gemstone.com
Voice 503-533-3365            Fax   503-629-8556
Gemstone Systems, Inc         http://www.gemstone.com

From kaffe@kaffe.org  Mon Oct 13 17:02:00 1997
From: kaffe@kaffe.org (Mark Huizer)
Date: Mon, 13 Oct 1997 18:02:00 +0200
Subject: JavaOS
In-Reply-To: <9710131412.AA07422@rkinc.com>; from "Aleksey Sudakov" on Mon, Oct 13, 1997 at 10:12:20AM -0400
References: <199710130657.IAA00386@diogenes.iit.upco.es> <9710131412.AA07422@rkinc.com>
Message-ID: <19971013180200.04495@pcnov032.win.tue.nl>

The wise Aleksey Sudakov produced the following lines:
> : 	Well, in the Computer Science Department at Utah University
> : you can find the answer to this discussion. They have already
> : developed a Java-OS using a library with provides you a toolkit to
> : build OS named OSkit upon and Intel ix86 architecture. I've used it
> : and it's really good. The OSKit code it's free and it's available in
> 
> It's totally amazing. The only bad things are that it's not available for  
> immediate download and it's not free. They say that it's free for  
> non-commercial use and IMHO it will affect acceptance of this totally cool  
> idea.
> 
Can you give me some URL or mailaddress so I can look at it? Sounds
cool, and yes, I'm not commercial planning to use it :-)

Mark

From kaffe@kaffe.org  Mon Oct 13 17:43:00 1997
From: kaffe@kaffe.org (Patrick Logan)
Date: Mon, 13 Oct 97 09:43 PDT
Subject: JavaOS
In-Reply-To: <9710131412.AA07422@rkinc.com>
References: <199710130657.IAA00386@diogenes.iit.upco.es>
 <9710131412.AA07422@rkinc.com>
Message-ID: <m0xKnaj-000DM5C@gharlane.gemstone.com>

>>>>> "Aleksey" == Aleksey Sudakov <zander@rkinc.com> writes:

    Aleksey> ...They say that it's free for non-commercial use...
    Aleksey> I wonder when free software become really free.

Why do you expect to make a profit from other people's labors? Isn't
it fair that, if you are going to charge money for something that you
didn't do, that you give a portion of that money to the ones who did
the work?

(Insert story of Henny Penny here.)

Look at it this way, giving money back to the ones who did the work
will keep *them* in business, so they can do *more* work.

Enough said in this list, please mail me personally about
disagreements.

-- 
Patrick Logan                 mailto:patrickl@gemstone.com
Voice 503-533-3365            Fax   503-629-8556
Gemstone Systems, Inc         http://www.gemstone.com

From kaffe@kaffe.org  Mon Oct 13 18:17:27 1997
From: kaffe@kaffe.org (Bartlee A. Anderson)
Date: Mon, 13 Oct 1997 12:17:27 -0500
Subject: JavaOS
References: <199710130657.IAA00386@diogenes.iit.upco.es>
Message-ID: <344257A7.1B4@switch.rockwell.com>

Luis,
I saw a reference to OSkit 5 months ago and its still not available
at the site you posted. What's the deal?
Bart.
Luis L. Fernandez wrote:
> 
> Hi,
> 
> 
>         I've listening your discussion about a new Java-OS and
> micro-kernels design....
> 
>         Well, in the Computer Science Department at Utah University
> you can find the answer to this discussion. They have already
> developed a Java-OS using a library with provides you a toolkit to
> build OS named OSkit upon and Intel ix86 architecture. I've used it
> and it's really good. The OSKit code it's free and it's available in
> 
> http://www.cs.utah.edu/projects/flux/oskit/
> 
> They are kernel hackers and they have redesigned the Mach micro-kernel
> to run fine (search about Fluke project).
> 
> I hope this message seems useful to somebody ;-)
> 
> Best Regards,
> 
>         Luis.
> 
> ---------
> Luis Lopez Fernandez
> e-mail: llf@computer.org

-- 
Bartlee A. Anderson     |  Sys. Test (ISDN-Commands-Trans Link-Tools)
Rockwell International  |                Electronic Commerce Division
300 Bauman Ct.          |                    banders@ecd.rockwell.com
MS 933-605      Opinions my own, not Rockwell's  VOICE (630) 227-8975
Wood Dale, IL 60191     |                        FAX   (630) 227-9771

From kaffe@kaffe.org  Mon Oct 13 18:24:11 1997
From: kaffe@kaffe.org (Luis L. Fernandez)
Date: Mon, 13 Oct 1997 19:24:11 +0200
Subject: JavaOS
In-Reply-To: <19971013180200.04495@pcnov032.win.tue.nl>
References: <199710130657.IAA00386@diogenes.iit.upco.es>
 <9710131412.AA07422@rkinc.com>
 <19971013180200.04495@pcnov032.win.tue.nl>
Message-ID: <199710131724.TAA02484@diogenes.iit.upco.es>

Mark Huizer writes:
 > > It's totally amazing. The only bad things are that it's not available for  
 > > immediate download and it's not free. They say that it's free for  
 > > non-commercial use and IMHO it will affect acceptance of this totally cool  
 > > idea.
 > > 
 > Can you give me some URL or mailaddress so I can look at it? Sounds
 > cool, and yes, I'm not commercial planning to use it :-)
 >

Sure, you can find the OSkit at this site:

	http://www.cs.utah.edu/projects/flux/oskit/

Enjoy!!

o-------------------------------o------------------------------------o
| Luis Lopez Fernandez          |  e-mail: luisl@iit.upco.es         |   
| Industrial Systems Division   |          llf@computer.org          |
o-------------------------------o------------------------------------o
|              Technological Research Institute                      |
|          Santa Cruz de Marcenado, 26 - Planta 1 -                  |
|	            28015 Madrid - SPAIN                             |
|      Tel: 34-1-5242800 (ext. 2706) - FAX: 34-1-5423176             |
o--------------------------------------------------------------------o

From kaffe@kaffe.org  Mon Oct 13 18:42:49 1997
From: kaffe@kaffe.org (Dan Lambright)
Date: Mon, 13 Oct 1997 13:42:49 -0400
Subject: JavaLobby and JavaOS
In-Reply-To: Your message of "Mon, 13 Oct 1997 07:17:31 EDT."
 <199710131117.HAA03790@radish.research.att.com>
Message-ID: <199710131742.NAA00507@postman.opengroup.org>

In message <199710131117.HAA03790@radish.research.att.com>Vijay Saraswat writes
>   X-Authentication-Warning: vega.soi.city.ac.uk: lp set sender to owner-kaffe
>@soi.city.ac.uk using -f
.
.
.
>I agree with Sean here. The notion of a strongly-typed,
>dynamically linked language, like Java, makes some fresh
>approaches possible. 
>
>To go a step beyond what Sean is saying: in fact it is possible
>(the "declarative bytecode verification" approach) to develop
>bytecode verifiers *systematically*, using a first-principles
>approach, from a constraint-based specification of the opcodes
>involved. One can translate an input byte code program, P,
>compositionally (byte-code for byte-code) into a constraint
>program, T(P) which "obviously" states the constraints on the
>run-time type-state of the Java Virtual Machine, which must hold
>for program execution to be type-safe. T(P) is "loop-free" and
>will take time proportional to its length, which is proportional
>to the length of P, to execute. If T(P) does not deadlock or
>produce "false", P is type-safe.

This aproach is taken by Penn State's active network's program for the PLAN 
language (which has its roots in ML). I have not seen it applied to Java by 
anyone. My group is considering it. A question I have concerns the expressive 
power a "loop free" language has to the developer. What can and cannot be 
described under such constraints? I would also be interested in what tradeoffs 
work well between load-time/verification time and expressability. How large 
can a classfile be?

>
>This approach differs slightly from the one that Sean and
>Javasoft have implemented. In their cases, there are separate
>*byte code verifiers*, usually largish (10-15 page) C programs,
>implementing a "data-flow" analysis, documented by 10-15 pages of
>informal English (I have not seen Sean's documentation yet),
>about whose correctness it is hard to say anything formal. 

I have not seen a formal desription of Java's bytecode verifier. I have 
emailed Sun and Princeton's SIP group- no reply.. anyone heard of efferts 
along these lines?

>In any case, the bottom line for this list should be, I think,
>that one can assume that the subproblem of verifying the
>type-safety of JVM-like bytecode is solved, with of course, one
>important caveat: native code. (A lot of the "core" classes in
>Java in fact contain natively implemented methods, and one has to
>use separate techniques for verifying them.) 

Dont download native code unless it is authenticated from someone you trust :-)
Software fault isolation is another avenue to investigate.

>I would not be too surprised if this approach (Java's approach)
>of doing "link-time" (i.e. first-instruction-execution-time)
>type-checking actually resulted in significant performance
>*improvements* over standard OS's. One thing to keep in mind is
>that Java's architecture is such that the bytecode verifier can
>in fact accept some input code *even if it is not type-safe in
>isolation* --- as long as it is type-safe when combined with the
>other classes that are already loaded. That is, there are some
>possibilities of doing inter-classfile optimizations (e.g.
>eliminating some classcasts generated by the compiler) in a
>typesafe way, that can produce even more performance benefits
>than are posible if one does "straight" Java verification, as
>done currently by Sun's Java systems.

How big a win would this be?

From kaffe@kaffe.org  Mon Oct 13 19:47:22 1997
From: kaffe@kaffe.org (David HM Spector)
Date: Mon, 13 Oct 1997 14:47:22 -0400
Subject: JavaOS
In-Reply-To: Your message of "Mon, 13 Oct 1997 19:24:11 +0200."
 <199710131724.TAA02484@diogenes.iit.upco.es>
Message-ID: <199710131847.OAA28502@wintermute.ny.zeitgeist.com>


From the OSKit web page:

The Current Release 

	The last release of the OS Toolkit is currently unavailable pending
	the preparation of a much enhanced, somewhat incompatible version due
	in June (Oh oh, we're into July already!). Contact Jay Lepreau if you
	have an urgent need for the previous release (but don't hold your
	breath).


I am interested in evaulating OSKit for a commercial project and
wanted to find out a ball-park for licensing fees; I had mailed and
called a number of times and got no response.  Not very encouraging.

Last week on the OSKit-user list the project coordinator Jay Lepreau
stated that they would try to put up a current snaphot after the
SOPS16 conference.

OSKit looks like a very kool piece of work, but the caveat from the
authors themselves seems to be a good idea.

As for the non-commcercial use clause, it would be interesting to see
what the real story actually is.  The OSKit docs clearly state that
their work is derived from other GPLd sources including
Linux... (note: not an opinion... just an observation...)

David

-- 
-------------------------------------------------------------------------------
David HM Spector                                         spector@zeitgeist.com
Network Design & Infrastructure Security                 voice: +1 212.579.8573
Amateur Radio: W2DHM (ex-N2BCA) (ARRL life member)       GridSquare: FN30AS
-.-. --- -. -. . -.-. -  .-- .. - ....  .- -- .- - . ..- .-.  .-. .- -.. .. ---
"New and stirring things are belittled because if they are not belittled, 
the humiliating question arises, 'Why then are you not taking part in them?'"
                                                        --H. G. Wells

From kaffe@kaffe.org  Mon Oct 13 20:55:13 1997
From: kaffe@kaffe.org (James A. Hillyerd)
Date: Mon, 13 Oct 1997 12:55:13 -0700 (PDT)
Subject: Utah OSKit/JavaOS
Message-ID: <Pine.LNX.3.95.971013125028.17003A-100000@animal.blarg.net>

Here is a link to a page that talks a little about Utah's Java Computer:

http://www.cs.utah.edu/projects/flux/oskit/javaos.html

It says it will be free for non-commercial use, and that a full release
was due out May 97.  I see that they are behind on almost all of their
deadlines.  (Must be a Microsoft train ground.)

I am curious, does Kaffe's license allow someone to port it, then sell it
for commercial use?  That doesn't seem right to me.  [Perhaps the JavaOS
just plain isn't available for commercial use, since they don't explicitly
say they are selling it.]

-james

[  James A. Hillyerd (JH2162) - james@blarg.net - Web Developer  ]
[  http://www.blarg.net/~james/   http://www.hyperglyphics.com/  ]
[ 1024/B11C3751 CA 1C B3 A9 07 2F 57 C9  91 F4 73 F2 19 A4 C5 88 ]

From kaffe@kaffe.org  Mon Oct 13 20:52:48 1997
From: kaffe@kaffe.org (Pramod B. S.)
Date: Tue, 14 Oct 1997 01:22:48 +0530 (IST)
Subject: JavaOS
In-Reply-To: <19971013180200.04495@pcnov032.win.tue.nl> from "Mark Huizer" at Oct 13, 97 06:02:00 pm
Message-ID: <9710131952.AA11370@rishi.serc.iisc.ernet.in>

> Can you give me some URL or mailaddress so I can look at it? Sounds
> cool, and yes, I'm not commercial planning to use it :-)
> 
> Mark
> 
> 

http://www.cs.utah.edu/projects/flux/
look under the section titled Software Distribution Pending.

Pramod

From kaffe@kaffe.org  Tue Oct 14 02:21:46 1997
From: kaffe@kaffe.org (Neil A. Carson)
Date: Mon, 13 Oct 1997 18:21:46 -0700
Subject: Utah OSKit/JavaOS
References: <Pine.LNX.3.95.971013125028.17003A-100000@animal.blarg.net>
Message-ID: <3442C92A.2781E494@causality.com>

James A. Hillyerd wrote:

> I am curious, does Kaffe's license allow someone to port it, then sell it
> for commercial use?  That doesn't seem right to me.  [Perhaps the JavaOS
> just plain isn't available for commercial use, since they don't explicitly
> say they are selling it.]

You'd be surprised at the number of packages that are written like this:
For example, the whole of the NetBSD OS, and anything else under the
Berkeley licence.

	Cheers,

	Neil

-- 
Neil A. Carson                         Marketing Director
E-mail: neil@causality.com             Causality Limited (London, UK)
Mobile: +44 (0)370 593183              Tel/Fax: +44 (0)181 930 7408
Sent from a 233MHz StrongARM NC!       Web: http://www.causality.com.

From kaffe@kaffe.org  Tue Oct 14 02:28:11 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Mon, 13 Oct 1997 19:28:11 -0600 (MDT)
Subject: Utah OSKit/JavaOS
In-Reply-To: <Pine.LNX.3.95.971013125028.17003A-100000@animal.blarg.net> from "James A. Hillyerd" at Oct 13, 97 12:55:13 pm
Message-ID: <199710140128.TAA09531@sal.cs.utah.edu>


 Hi,

I'm the guy from the University of Utah who did the Kaffe to OS Kit
port. Unfortunately, I can't speak for my employer (Jay Lepreau), but
the situation as I see it is as follows.

We're behind our release schedule mainly for reasons of licensing which 
had to be resolved with our funders and the university. I cannot be more
specific at this point. Jay has told us that we will have at least a 
snapshot out by the end of next week. The license will allow free use 
for (at least) non-commercial purposes. The GNU-related/derived 
parts will obviously be distributed under the GNU license. 

As for the Kaffe port, it was done using an older version of the OS Kit
and using an older version of Kaffe (0.8.1). Although I don't currently
work on JavaOS, I want to make the port publicly available. As I'm doing
work-for-hire for the university, it won't be up to me to decide under
which license this will be distributed, but it will certainly not be in 
disagreement with Kaffe's license - which is a BSD-style license.
Before I make the port available, I would like to update to 0.9.2.

It is our hope that the OS Kit and the Kaffe port will spark research
in the areas that were discussed earlier in this thread. 

	- Godmar (gback@cs.utah.edu)

From kaffe@kaffe.org  Tue Oct 14 09:00:43 1997
From: kaffe@kaffe.org (Luis L. Fernandez)
Date: Tue, 14 Oct 1997 10:00:43 +0200
Subject: JavaOS
In-Reply-To: <344257A7.1B4@switch.rockwell.com>
References: <199710130657.IAA00386@diogenes.iit.upco.es>
 <344257A7.1B4@switch.rockwell.com>
Message-ID: <199710140800.KAA04485@diogenes.iit.upco.es>

Bartlee A. Anderson writes:
 > Luis,
 > I saw a reference to OSkit 5 months ago and its still not available
 > at the site you posted. What's the deal?

You are right, but this site give you a link to the person who is
responsible for OSKit project (Jay Lepreau). If you need the last
version of this library, please contact with him
(lepreau@cs.uath.edu).

Anyway if you have an urgent need I'm going to leave the oskit-0.60
release, the last one I have used, in the site:

ftp://nautilus.gsyc.inf.uc3m.es/pub/

Regards from Spain, Bart. If you need something else please, let me
know!!.

	Luis.

P.S: We have used OSKit to develop a "new" minimal micro-kernel from
the scratch. See "http://www.iit.upco.es/~luisl/advice.html" for more
info about this one. We have reduced the development time for six
months by using this toolkit. In my point, this one is a great effort
to lighten the OS development task.


o-------------------------------o------------------------------------o
| Luis Lopez Fernandez          |  e-mail: luisl@iit.upco.es         |   
| Industrial Systems Division   |          llf@computer.org          |
|                               |  http://www.iit.upco.es/~luisl     |      
o-------------------------------o------------------------------------o
|              Technological Research Institute                      |
|          Santa Cruz de Marcenado, 26 - Planta 1 -                  |
|	            28015 Madrid - SPAIN                             |
|      Tel: 34-1-5242800 (ext. 2706) - FAX: 34-1-5423176             |
o--------------------------------------------------------------------o

From kaffe@kaffe.org  Tue Oct 14 16:44:59 1997
From: kaffe@kaffe.org (Peter Michael Bertelsen)
Date: Tue, 14 Oct 1997 17:44:59 +0200
Subject: Formal JVM specification
In-Reply-To: <199710131742.NAA00507@postman.opengroup.org> (message from Dan
 Lambright on Mon, 13 Oct 1997 13:42:49 -0400)
Message-ID: <199710141544.RAA10521@ellemose.dina.kvl.dk>

Dan Lambright wrote:
> I have not seen a formal desription of Java's bytecode verifier. I have 
> emailed Sun and Princeton's SIP group- no reply.. anyone heard of efferts 
> along these lines?

I've written a report, `Semantics of Java Byte Code', giving a formal
description of the JVM and the bytecode instructions.  The JVM is
modelled as an infinite state machine, and each byte code instruction
is described by a transition rule, describing the state change and the
conditions under which execution succeeds.  The report also lists a
number of errors and ambiguities in Sun's specification of the JVM.

This might be of interest to you, although some aspects of class file
verification are not covered, e.g. checking that the file has the
correct structure.  The report is available at:

  ftp://ftp.dina.kvl.dk/pub/Staff/Peter.Bertelsen/jvm-semantics.ps.gz

Also check out the `Defensive JVM' project at Computational Logic:

  http://www.cli.com/software/djvm/

-Peter Bertelsen

From kaffe@kaffe.org  Tue Oct 14 18:00:21 1997
From: kaffe@kaffe.org (Bartlee A. Anderson)
Date: Tue, 14 Oct 1997 12:00:21 -0500
Subject: JavaOS
References: <199710130657.IAA00386@diogenes.iit.upco.es>
 <344257A7.1B4@switch.rockwell.com> <199710140800.KAA04485@diogenes.iit.upco.es>
Message-ID: <3443A525.7EE6@switch.rockwell.com>

Getting it now,
Thanks,
	Bart

Luis L. Fernandez wrote:
> 
> Bartlee A. Anderson writes:
>  > Luis,
>  > I saw a reference to OSkit 5 months ago and its still not available
>  > at the site you posted. What's the deal?
> 
> You are right, but this site give you a link to the person who is
> responsible for OSKit project (Jay Lepreau). If you need the last
> version of this library, please contact with him
> (lepreau@cs.uath.edu).
> 
> Anyway if you have an urgent need I'm going to leave the oskit-0.60
> release, the last one I have used, in the site:
> 
> ftp://nautilus.gsyc.inf.uc3m.es/pub/
> 
> Regards from Spain, Bart. If you need something else please, let me
> know!!.
> 
>         Luis.
> 
> P.S: We have used OSKit to develop a "new" minimal micro-kernel from
> the scratch. See "http://www.iit.upco.es/~luisl/advice.html" for more
> info about this one. We have reduced the development time for six
> months by using this toolkit. In my point, this one is a great effort
> to lighten the OS development task.
> 
> o-------------------------------o------------------------------------o
> | Luis Lopez Fernandez          |  e-mail: luisl@iit.upco.es         |
> | Industrial Systems Division   |          llf@computer.org          |
> |                               |  http://www.iit.upco.es/~luisl     |
> o-------------------------------o------------------------------------o
> |              Technological Research Institute                      |
> |          Santa Cruz de Marcenado, 26 - Planta 1 -                  |
> |                   28015 Madrid - SPAIN                             |
> |      Tel: 34-1-5242800 (ext. 2706) - FAX: 34-1-5423176             |
> o--------------------------------------------------------------------o

-- 
Bartlee A. Anderson     |  Sys. Test (ISDN-Commands-Trans Link-Tools)
Rockwell International  |                Electronic Commerce Division
300 Bauman Ct.          |                    banders@ecd.rockwell.com
MS 933-605      Opinions my own, not Rockwell's  VOICE (630) 227-8975
Wood Dale, IL 60191     |                        FAX   (630) 227-9771

From kaffe@kaffe.org  Wed Oct 15 10:17:22 1997
From: kaffe@kaffe.org (Malek Shabou)
Date: Wed, 15 Oct 1997 10:17:22 +0100
Subject: Pb with kaffe on win95
Message-ID: <199710150917.KAA29987@saturne.inria.fr>

 Hello
 I've compiled kaffe on a win95 succefuly, but when i try to run HelloWorldApp an error message apers:

kaffe -v -verbosejit HelloWorldApp

Loading class 'java/lang/Throwable'.
Loading class 'java/lang/IllegalArgumentException'.
Loading class 'java/lang/RuntimeException'.
Loading class 'java/lang/IllegalThreadStateException'.
Loading class 'java/lang/InterruptedException'.
Loading class 'java/lang/NoSuchMethodError'.
Loading class 'java/lang/IncompatibleClassChangeError'.
Loading class 'java/lang/LinkageError'.
Loading class 'java/lang/Error'.
Loading class 'java/lang/SecurityManager'.
Loading class 'java/lang/StringBuffer'.
Loading class 'java/lang/ThreadDeath'.
Loading class 'java/lang/ThreadGroup'.
Loading class 'java/lang/CloneNotSupportedException'.
Loading class 'java/lang/Integer'.
Loading class 'java/lang/Number'.
Loading class 'java/io/PrintStream'.
Loading class 'java/io/FilterOutputStream'.
Loading class 'java/io/OutputStream'.
Loading class 'java/lang/NullPointerException'.
Loading class 'sun/misc/VM'.
<JIT: java/lang/ThreadGroup.add(Ljava/lang/Thread;)V time 0ms (0ms) @ 0x4b82818>


(C:\KAFFE\BIN\KAFFE.EXE 1000) In cygwin_except_handler



  can any one help me?


A+


-- 
Malek Shabou				E-mail: Malek.Shabou@inria.fr
NIC France
INRIA-Rocquencourt Domaine de Voluceau
B.P.105 78153 Le Chesnay Cedex

From kaffe@kaffe.org  Wed Oct 15 16:52:09 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Thu, 16 Oct 1997 01:52:09 +1000 (EST)
Subject: NoSuchMethodError
Message-ID: <199710151552.BAA12624@zen.quick.com.au>

I was trying to get Servlets running under appache on a SunOS system
today using kaffe as the JVM.  The same apache config etc on a Solaris
system using jdk1.1.4 worked fine so I know everything other than
kaffe is cool (I tried 0.8.4,0.9.1,0.9.2).

Anyway, whenever I try to run a servlet via the jserv_mod[ule] kaffe
just says:

NoSuchMethodError

which is obviously sub-optimal from a debugging point of view.

I've not looked myself, but would it be difficult to modify things so
that one learned which method of which class was not found?

I've not yet tried using other servlet engines with kaffe - has anyone
had success with kaffe running servlets?

--sjg

From kaffe@kaffe.org  Wed Oct 15 21:48:28 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Wed, 15 Oct 1997 14:48:28 -0600 (MDT)
Subject: NoSuchMethodError
In-Reply-To: <199710151552.BAA12624@zen.quick.com.au> from "Simon J. Gerraty" at Oct 16, 97 01:52:09 am
Message-ID: <199710152048.OAA24151@lal.cs.utah.edu>

'Simon J. Gerraty' wrote:
> I've not looked myself, but would it be difficult to modify things so
> that one learned which method of which class was not found?

I've made the changes myself, but my local copy of Kaffe is whacked in
a number of respects, and you don't want to see my diff.  The actual
changes are pretty simple.  In kaffe/kaffevm/, grep for
'NoSuchMethodError', in many of the cases (e.g., in support.c) you'll
want to change those instances to 'NoSuchMethodErrorMsg(method_name)',
where "method_name" is some C string that describes the method.
For a simple hack, just use the method_name passed into the function,
in other places you may want to construct a more informative message
(e.g., "no such static method", or something.)  

Now you just need to define NoSuchMethodErrorMsg(M), which you'll put
in errors.h:
#define NoSuchMethodErrorMsg(M)  NEW_LANG_EXCEPTION_MESSAGE(NoSuchMethodError, M)

Hope this helps more than a diff....

-Pat


From kaffe@kaffe.org  Thu Oct 16 04:51:01 1997
From: kaffe@kaffe.org (Michael Madore)
Date: Wed, 15 Oct 1997 20:51:01 -0700
Subject: no math in shared library path
Message-ID: <3.0.1.32.19971015205101.006aaf88@sunshine>

Hi,

I'm running Kaffe 0.9.2 on Debian Linux (2.0.30 kernel).  When I attempt to
use the BigInteger class in my program I get the following error:

   Java.lang.UnsatisfiedLinkError:   no math in shared library path

I'm using Oracle's thin JDBC drivers, which apparently make use of this
class.  The error occurs if I use the class in a simple program without
JDBC also.  If I run the compiled program on an NT machine, I don't have
any problems.  Is this caused by my Linux installation, or by Kaffe?

TIA

Mike Madore

From kaffe@kaffe.org  Thu Oct 16 09:04:36 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Thu, 16 Oct 1997 18:04:36 +1000 (EST)
Subject: NoSuchMethodError
References: <199710152048.OAA24151@lal.cs.utah.edu>
Message-ID: <199710160804.SAA16176@gate.quick.com.au>

>'Simon J. Gerraty' wrote:
>> I've not looked myself, but would it be difficult to modify things so
>> that one learned which method of which class was not found?

>I've made the changes myself, but my local copy of Kaffe is whacked in

Thanks, I've made the appropriate changes and re-installed kaffe
I could not of course get it ti generate that error... :-)
Might have been kaffe-0.8.4 doing that :-(, anyway the patches
seem a neat addition - though I'll need to test what the result
looks like.

The good news is that after getting no where via jserv, I ran:

java sun.servlet.http.HttpServer

with appropriate CLASSPATH (java is my java_wrapper which runs kaffe) 
and using lynx, I can get SnoopServlet running.  
So presumably the problem is related to jserv.
I'll try soem other servlet engines and see how it goes.

--sjg




-- 
Simon J. Gerraty        <sjg@quick.com.au>

#include <disclaimer>   /* imagine something _very_ witty here */

From kaffe@kaffe.org  Thu Oct 16 10:13:38 1997
From: kaffe@kaffe.org (Malek Shabou)
Date: Thu, 16 Oct 1997 10:13:38 +0100
Subject: kaffe 0.9.2 with win95
Message-ID: <199710160913.KAA04563@saturne.inria.fr>

 Hello,
 i've compiled kaffe 0.9.2 under win95 envirenment, and when i trie to execute 
HelloWorldApp or any other application i get this:
Loading class 'java/lang/Throwable'.
Loading class 'java/lang/IllegalArgumentException'.
Loading class 'java/lang/RuntimeException'.
Loading class 'java/lang/IllegalThreadStateException'.
Loading class 'java/lang/InterruptedException'.
Loading class 'java/lang/NoSuchMethodError'.
Loading class 'java/lang/IncompatibleClassChangeError'.
Loading class 'java/lang/LinkageError'.
Loading class 'java/lang/Error'.
Loading class 'java/lang/SecurityManager'.
Loading class 'java/lang/StringBuffer'.
Loading class 'java/lang/ThreadDeath'.
Loading class 'java/lang/ThreadGroup'.
Loading class 'java/lang/CloneNotSupportedException'.
Loading class 'java/lang/Integer'.
Loading class 'java/lang/Number'.
Loading class 'java/io/PrintStream'.
Loading class 'java/io/FilterOutputStream'.
Loading class 'java/io/OutputStream'.
Loading class 'java/lang/NullPointerException'.
Loading class 'sun/misc/VM'.
<JIT: java/lang/ThreadGroup.add(Ljava/lang/Thread;)V time 0ms (0ms) @ 0x4b82818>


(C:\KAFFE\BIN\KAFFE.EXE 1000) In cygwin_except_handler





can any help me,

Cordialement,

-- 
Malek Shabou				E-mail: Malek.Shabou@inria.fr
NIC France
INRIA-Rocquencourt Domaine de Voluceau
B.P.105 78153 Le Chesnay Cedex


From kaffe@kaffe.org  Thu Oct 16 18:46:21 1997
From: kaffe@kaffe.org (James Cooper)
Date: Thu, 16 Oct 1997 10:46:21 -0700 (PDT)
Subject: solaris 2.5 + socket = core dump
Message-ID: <Pine.GSO.3.95.971016104047.19310B-100000@caustik.organic.com>


Hi,

I'm trying to get Apache's JServ running under kaffe
(http://java.apache.org).  JServ is a runtime environment for servlets
that work with apache.  It doesn't use any native methods of its own.

A little info about my setup:
  Sparc 5 - Solaris 2.5
  kaffe 0.9.1
  classes.zip from the Sun JDK 1.1.2

HelloWorldApp runs fine, as does the javac wrapper, so I was psyched.  I
configured JServ to use the kaffe VM instead of the Sun VM, and fired it
up.

JServ starts up fine, and begins listening on a socket.  However, when an
incoming connection comes in it dumps core.  Here's the stack trace from
gdb:

(gdb) where
#0  0xef6f4140 in _kill ()
#1  0xef6b965c in abort ()
#2  0xef4a2a1c in java_net_PlainSocketImpl_socketGetOption (this=0x276490, 
    val=15) at ./java.net/PlainSocketImpl.c:242
#3  0x29f140 in _end ()
#4  0x2a32c4 in _end ()
#5  0x2a7164 in _end ()
#6  0x271cc4 in _end ()
#7  0xef754420 in do_execute_java_method (ee=0xef774528, obj=0x14f058, 
    method_name=0x14f058 "", signature=0xef774528 "()V", mb=0x130890, 
    isStaticCall=0) at support.c:83
#8  0xef751264 in firstStartThread () at thread.c:206
#9  0xef752c30 in reschedule () at thread-internal.c:497
#10 0xc in ?? ()


I know very little about gdb, so that didn't mean much to me, but I
thought it might be helpful to someone here.

I've noticed that other folks on the list are successfully running servers
under kaffe, so I imagine it's something funky with my setup.  If anyone
has any suggestions on how to better diagnose the cause of this problem
I'd be appreciative.

Thanks

-- James

--------------------------------------------------------------------------------
                              James Paul Cooper
                      Organic Online - pixel@organic.com
                          ......Know Future......

From kaffe@kaffe.org  Fri Oct 17 05:44:26 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Fri, 17 Oct 1997 14:44:26 +1000 (EST)
Subject: kaffe can handle servlets
Message-ID: <199710170444.OAA29667@gate.quick.com.au>

FYI, a trivial mod to apache's mod_jserv.c (replace "rb+" with "r+b")
gets servlets running on SunOS 4.1.4 with kaffe-0.9.2.

When I try this on NetBSD with kaffe-0.9.1 I just get:

[Fri Oct 17 14:20:41 1997] access to /share/java/JSDK1.0.1/servlets/DateServlet failed for xx.xx.xx.xx, reason: Premature end of script headers

and of course I've not got 0.9.2 working yet.

--sjg


From kaffe@kaffe.org  Fri Oct 17 05:41:00 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Fri, 17 Oct 1997 14:41:00 +1000 (EST)
Subject: kaffe-0.9.2: core dumps on NetBSD/i386
Message-ID: <199710170441.OAA29603@gate.quick.com.au>


kaffe-0.9.2 builds ok on NetBSD/i386 but if I try and
run java foo which should produce 
java.lang.ClassNotFoundException: foo
I get
Illegal instruction (core dumped)

The stack trace looks a bit odd, so I tried:

sjg:202$ ulimit -s 32768
sjg:203$ java foo
[wait....,wait...,....]
/share: write failed, file system is full
Illegal instruction 

I would guess that something is recursing itself to death.

--sjg

From kaffe@kaffe.org  Fri Oct 17 05:25:16 1997
From: kaffe@kaffe.org (PookMan)
Date: Fri, 17 Oct 1997 00:25:16 -0400 (EDT)
Subject: Kaffe or coffee ?
Message-ID: <Pine.SUN.3.91.971017002432.8906B-100000@lisa.cs.mcgill.ca>


 I am sorry if this sounds like a stupid question but In the README file, 
we read this :

For a Kaffe system you need the following parts:

 kaffe-0.9.1.tgz                        Virtual machine and all bits
                                        provided by tjwassoc.co.uk
 kaffe-0.9.1-package-javasoft.com.tgz   Sun's class libraries.


If you want to use graphics you'll need one of the following:

 kaffe-0.9.1-package-biss-net.com.tgz   AWT graphics support.
 kaffe-0.9.1-package-epfl.ch.tgz        Alternative graphics support.


  I simply want to know where we can get those packages ?? 
At the corresponding site ? Where exactly ? ftp address ?

 THank you for your help

From kaffe@kaffe.org  Fri Oct 17 08:43:22 1997
From: kaffe@kaffe.org (PookMan)
Date: Fri, 17 Oct 1997 03:43:22 -0400 (EDT)
Subject: Kaffe or coffee ? (fwd)
Message-ID: <Pine.SUN.3.91.971017034047.16050A-100000@lisa.cs.mcgill.ca>


 Please forget my previous question and focus on this one :)
 
I am using BSDI 2.1 and I am trying to install Kaffe

 I get the following error at the compiling

 true --def ./libkaffe_bissawt.def --output-exp lib.exp --output-lib 
libkaffe_bissawt.a --dllname libkaffe_bissawt.a
touch lib.exp
ar cr libkaffe_bissawt.a.0.87 biss_awt_kernel_NativeLib.o  color.o  
display.o  font.o  graphics.o  image.o  window.o
ar: warning: biss_awt_kernel_NativeLib.o truncated to biss_awt_kernel
ranlib libkaffe_bissawt.a.0.87
rm -f libkaffe_bissawt.a
ln -s libkaffe_bissawt.a.0.87 libkaffe_bissawt.a
gcc -g -O -DNO_SHARED_LIBRARIES -I. -I. -I./../../config -I../../config 
-I../../include -I./../../include -DTRANSLATOR -I./jit -DKVER=\"0.91\" 
-c  ./jit/funcs.c
funcs.c:8805:Unknown pseudo-op:  '.global'
funcs.c:8805:Rest of line ignored. 1st junk character valued 95 (_).
*** Error code 1

Stop.
*** Error code 1

Stop.
*** Error code 1

Stop.
 
 Does anyone has any idea on how to solve this problem ?

 I would really appreciate any tip from you ...

 thank you

From kaffe@kaffe.org  Fri Oct 17 08:40:07 1997
From: kaffe@kaffe.org (Artur Biesiadowski)
Date: Fri, 17 Oct 1997 09:40:07 +0200 (MET DST)
Subject: Java inside gcc ?
Message-ID: <Pine.GSO.3.96L01AT.971016195031.2682A-100000@sunrise>



At what stage is project of integrating java/kaffe with gcc ?

Artur

From kaffe@kaffe.org  Fri Oct 17 11:50:16 1997
From: kaffe@kaffe.org (Fathi.DJEBBARI)
Date: Fri, 17 Oct 1997 11:50:16 +0100
Subject: Kaffe or coffee ? (fwd)
Message-ID: <199710171050.LAA02589@bcv64s3e.vz.cit.alcatel.fr>


pooky wrote 
> From kaffe-approval@soi.city.ac.uk Fri Oct 17 12:42 MET 1997
> X-Authentication-Warning: vega.soi.city.ac.uk: lp set sender to owner-kaffe@soi.city.ac.uk using -f
> Date: Fri, 17 Oct 1997 03:43:22 -0400 (EDT)
> From: PookMan <pooky@cs.mcgill.ca>
> To: kaffe@kaffe.org
> Subject: Kaffe or coffee ? (fwd)
> Mime-Version: 1.0
> Sender: owner-kaffe@soi.city.ac.uk
> Reply-To: kaffe@kaffe.org
> Content-Type> : > TEXT/PLAIN> ; > charset=US-ASCII> 
> Content-Length: 1056
> 
> 
>  Please forget my previous question and focus on this one :)
>  
> I am using BSDI 2.1 and I am trying to install Kaffe
> 
>  I get the following error at the compiling
> 
>  true --def ./libkaffe_bissawt.def --output-exp lib.exp --output-lib 
> libkaffe_bissawt.a --dllname libkaffe_bissawt.a
> touch lib.exp
> ar cr libkaffe_bissawt.a.0.87 biss_awt_kernel_NativeLib.o  color.o  
> display.o  font.o  graphics.o  image.o  window.o
> ar: warning: biss_awt_kernel_NativeLib.o truncated to biss_awt_kernel
> ranlib libkaffe_bissawt.a.0.87
> rm -f libkaffe_bissawt.a
> ln -s libkaffe_bissawt.a.0.87 libkaffe_bissawt.a
> gcc -g -O -DNO_SHARED_LIBRARIES -I. -I. -I./../../config -I../../config 
> -I../../include -I./../../include -DTRANSLATOR -I./jit -DKVER=\"0.91\" 
> -c  ./jit/funcs.c
> funcs.c:8805:Unknown pseudo-op:  '.global'
> funcs.c:8805:Rest of line ignored. 1st junk character valued 95 (_).
> *** Error code 1
> 
> Stop.
> *** Error code 1
> 
> Stop.
> *** Error code 1
> 
> Stop.
>  
>  Does anyone has any idea on how to solve this problem ?
> 
>  I would really appreciate any tip from you ...
> 
>  thank you
> 
> 
I am not really sure of my answer, but it seems to me that bsdi assembler
work with globl lalel instead of global.
Bye
Fathi

From kaffe@kaffe.org  Fri Oct 17 13:54:02 1997
From: kaffe@kaffe.org (Martin Lanzen)
Date: Fri, 17 Oct 1997 14:54:02 +0200 (MET DST)
Subject: Kaffe or coffee ?
In-Reply-To: <Pine.SUN.3.91.971017002432.8906B-100000@lisa.cs.mcgill.ca>
Message-ID: <199710171254.OAA27860@puh>

On 17 Okt , PookMan wrote:
> 
>  I am sorry if this sounds like a stupid question but In the README file, 
> we read this :
> 
> For a Kaffe system you need the following parts:
> 
>  kaffe-0.9.1.tgz                        Virtual machine and all bits
>                                         provided by tjwassoc.co.uk
>  kaffe-0.9.1-package-javasoft.com.tgz   Sun's class libraries.
> 
> 
> If you want to use graphics you'll need one of the following:
> 
>  kaffe-0.9.1-package-biss-net.com.tgz   AWT graphics support.
>  kaffe-0.9.1-package-epfl.ch.tgz        Alternative graphics support.
> 
> 
>   I simply want to know where we can get those packages ?? 
> At the corresponding site ? Where exactly ? ftp address ?
> 
>  THank you for your help
> 
I found these files at:
 
http://www.oasis.leo.org/java/machines/kaffe/00-index.html
-- 
                                                           
/Martin                                        

From kaffe@kaffe.org  Fri Oct 17 18:46:19 1997
From: kaffe@kaffe.org (Vik Sohal)
Date: Fri, 17 Oct 1997 10:46:19 -0700
Subject: Location of packages for kaffe
Message-ID: <2.2.32.19971017174619.00bc7a8c@wizard.lynx.com>


You can get the BISS file from:

http://btp1da.phy.uni-bayreuth.de/~werner/FreeBSD/fbsd_pub_ftp/distfiles/

The epfl file can be gotten from:

http://gd.tuwien.ac.at/languages/java/kaffe/ghindex.shtml

Vik
***************************************************************
* Vik Sohal                     *  email: vik@lynx.com        *
* Lynx Real-Time Systems        *  voice: (408) 879-3900 x119 *
* 2239 Samaritan Drive          *    fax: (408) 879-3920      *
* San Jose, CA 95124            *                             *
* ------------------------------*-----------------------------*
* cute msg: "We just turned it on, and it's I.Q. is already   *
*             in the thousands!"                              *
***************************************************************

From kaffe@kaffe.org  Fri Oct 17 20:42:20 1997
From: kaffe@kaffe.org (Per Bothner)
Date: Fri, 17 Oct 1997 12:42:20 -0700
Subject: Java inside gcc ?
In-Reply-To: Your message of "Fri, 17 Oct 1997 09:40:07 +0200."
 <Pine.GSO.3.96L01AT.971016195031.2682A-100000@sunrise>
Message-ID: <199710171941.MAA29724@cygnus.com>

> At what stage is project of integrating java/kaffe with gcc ?

Assuming you mean:  Integrating gcc-compiled Java code with Kaffe,
then:  Fairly far.  The major outstanding issue is exception handling.

	--Per Bothner
Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner

From kaffe@kaffe.org  Fri Oct 17 23:55:20 1997
From: kaffe@kaffe.org (Per Bothner)
Date: Fri, 17 Oct 1997 15:55:20 -0700
Subject: Java inside gcc ?
In-Reply-To: Your message of "Fri, 17 Oct 1997 17:20:43 CDT."
 <199710172220.RAA16096@compound.east.sun.com>
Message-ID: <199710172254.PAA08199@cygnus.com>

Tony Kimball <alk@pobox.com> writes:

> Will it be part of egcs?

That is one possibility, but it has not been decided.
(It partly depends on the egcs release plans.)

	--Per Bothner
Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner

From kaffe@kaffe.org  Sat Oct 18 02:28:10 1997
From: kaffe@kaffe.org (David Morton)
Date: Fri, 17 Oct 1997 20:28:10 -0500 (CDT)
Subject: Java inside gcc ?
In-Reply-To: <199710171941.MAA29724@cygnus.com>
Message-ID: <XFMail.971017202810.mortonda@dgm.cadvantage.com>

I'm really getting anxious... :)

On 17-Oct-97 Per Bothner wrote:
>> At what stage is project of integrating java/kaffe with gcc ?
> 
> Assuming you mean:  Integrating gcc-compiled Java code with Kaffe,
> then:  Fairly far.  The major outstanding issue is exception handling.
> 
>       --Per Bothner
> Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner

--------------------------------------------------------
David Morton         See my webpage for my public pgp key                 
E-Mail: mortonda@esuvm.emporia.edu           
WWW: http://thinker.emporia.edu/~mortonda/home.html
Date: 17-Oct-97 Time: 20:27:39  CST -600 GMT        

From kaffe@kaffe.org  Sat Oct 18 11:35:46 1997
From: kaffe@kaffe.org (Preben Randhol)
Date: 18 Oct 1997 12:35:46 +0200
Subject: Graphics
Message-ID: <jclk9fbmk8t.fsf@iq.pvv.ntnu.no>

I've tried to run a small java-program in X Windows. It works with the
ordinary java interpreter, but when I try to run it with kaffe there is just
a lot of error-msgs. It is only a small window with a button on it. Nothing
fancy. Kaffe works great with programs that only outputs text to STDOUT
though. Is there some thing I need ? I have installed the biss and eplch packages.

Thanks in advance

:info
Kaffe 0.9.1
Platform i586 with RedHat 4.2 Linux
Kernel 2.0.30

-- 
Preben Randhol                    | Soldiers are followers to the
Tlf:   73940929    Arb: 73594076  | extreme.
Epost: randhol@pvv.org            |
http://www.pvv.org/~randhol/      |     

From kaffe@kaffe.org  Sat Oct 18 21:22:57 1997
From: kaffe@kaffe.org (Corey Minyard)
Date: 18 Oct 1997 15:22:57 -0500
Subject: Java inside gcc ?
In-Reply-To: Per Bothner's message of Fri, 17 Oct 1997 12:42:20 -0700
References: <199710171941.MAA29724@cygnus.com>
Message-ID: <m2iuuuzuqm.fsf@wf-rch.cirr.com>

Per Bothner <bothner@cygnus.com> writes:

> 
> > At what stage is project of integrating java/kaffe with gcc ?
> 
> Assuming you mean:  Integrating gcc-compiled Java code with Kaffe,
> then:  Fairly far.  The major outstanding issue is exception handling.
> 
> 	--Per Bothner
> Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner
> 

Just a thought, you might want to talk to the GNAT people (the GNU
version of Ada supported by Ada Core Technologies, if you don't know)
about exception handling.  Ada and Java exception handling are pretty
close semantically and they might have useful stuff for you.

-- 
Corey Minyard               Internet:  minyard@acm.org
  Work: minyard@nortel.ca       UUCP:  minyard@wf-rch.cirr.com

From kaffe@kaffe.org  Mon Oct 20 07:54:35 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Mon, 20 Oct 1997 16:54:35 +1000 (EST)
Subject: kaffe can handle servlets
References: <199710170444.OAA29667@gate.quick.com.au>
Message-ID: <199710200654.QAA05675@gate.quick.com.au>

>FYI, a trivial mod to apache's mod_jserv.c (replace "rb+" with "r+b")
>gets servlets running on SunOS 4.1.4 with kaffe-0.9.2.

>When I try this on NetBSD with kaffe-0.9.1 I just get:

>[Fri Oct 17 14:20:41 1997] access to /share/java/JSDK1.0.1/servlets/DateServlet failed for xx.xx.xx.xx, reason: Premature end of script headers

The solution to this was the following patch:

*** kaffe-0.9.1/packages/tjwassoc.co.uk/APInet/lib/java.net/PlainSocketImpl.c.~1~    Fri Jun 27 01:02:10 1997
--- kaffe-0.9.1/packages/tjwassoc.co.uk/APInet/lib/java.net/PlainSocketImpl.c        Sun Jul 27 10:26:50 1997
***************
*** 233,243 ****
  void
  java_net_PlainSocketImpl_socketSetOption(struct Hjava_net_PlainSocketImpl* this, jint v1, jint /* bool */ v2, struct Hjava_lang_Object* v3)
  {
!       abort();
  }
  
  jint
  java_net_PlainSocketImpl_socketGetOption(struct Hjava_net_PlainSocketImpl* this, jint val)
  {
!       abort();
  }
--- 233,245 ----
  void
  java_net_PlainSocketImpl_socketSetOption(struct Hjava_net_PlainSocketImpl* this, jint v1, jint /* bool */ v2, struct Hjava_lang_Object* v3)
  {
!       /* abort(); /* no thanks --sjg */
  }
  
  jint
  java_net_PlainSocketImpl_socketGetOption(struct Hjava_net_PlainSocketImpl* this, jint val)
  {
!       /* abort(); /* no thanks --sjg */
!       val = 0;
!       return 0;
  }

-- 
Simon J. Gerraty        <sjg@quick.com.au>

#include <disclaimer>   /* imagine something _very_ witty here */

From kaffe@kaffe.org  Mon Oct 20 10:52:59 1997
From: kaffe@kaffe.org (Iso-H)
Date: Mon, 20 Oct 1997 12:52:59 +0300 (EET DST)
Subject: Solaris 2.5 + kaffe = core dump
In-Reply-To: <Pine.GSO.3.95.971016104047.19310B-100000@caustik.organic.com>
Message-ID: <Pine.LNX.3.95q.971020124525.9063A-100000@linray.ts.ray.fi>

On Thu, 16 Oct 1997, James Cooper wrote:

> 
> Hi,
> 
> I'm trying to get Apache's JServ running under kaffe
> (http://java.apache.org).  JServ is a runtime environment for servlets
> that work with apache.  It doesn't use any native methods of its own.
> 
> A little info about my setup:
>   Sparc 5 - Solaris 2.5
>   kaffe 0.9.1
>   classes.zip from the Sun JDK 1.1.2
> 

  Did you compile kaffe as dynamic? I can't get kaffe to work
  (under Solaris) if it is compiled as dynamic; it just dumps 
  core all the time!

  But if I use configure's option "--enable-staticlib" then kaffe
  works, but "KaffeVerifyBug" test gives this:

  java.lang.UnsatisfiedLinkError: no net in shared library path
        at java/lang/Throwable.<init>(line unknown, pc 16bf2c) 
        at java/lang/Error.<init>(line unknown, pc 18b654) 
        at java/lang/LinkageError.<init>(line unknown, pc 18b5dc) 
        at java/lang/UnsatisfiedLinkError.<init>(line unknown, pc 18b564) 
        at java/lang/Runtime.loadLibrary(line unknown, pc 16fbc0) 
        at java/lang/System.loadLibrary(line unknown, pc 18adf0) 
        at java/net/InetAddress.<clinit>(line unknown, pc 19219c) 
        at java/net/Socket.<init>(97)
        at KaffeVerifyBug.doit(21)
        at KaffeVerifyBug.main(9)


Iso-H

--

From kaffe@kaffe.org  Mon Oct 20 21:22:13 1997
From: kaffe@kaffe.org (Amitabh Arora)
Date: Mon, 20 Oct 1997 13:22:13 -0700
Subject: Binaries for Sun4.1 and HP 9.05 OSs
Message-ID: <01BCDD5B.2DFF26B0@amitabh@aspec.com>

Does anyone have Kaffe running on Sun 4.1.3 and HP-UX 9.05?
If yes, then please let me know how to download the binaries?

Thanks. -amitabh arora

From kaffe@kaffe.org  Mon Oct 20 23:10:10 1997
From: kaffe@kaffe.org (Alexandre Oliva)
Date: 20 Oct 1997 20:10:10 -0200
Subject: kaffe 0.9.* on IRIX 5.2: threads.s
Message-ID: <ord8l0gk71.fsf@grupiara.dcc.unicamp.br>

--Multipart_Mon_Oct_20_20:10:09_1997-1
Content-Type: text/plain; charset=US-ASCII

Hi there!

I've been discussing with Ian Lance Taylor, GNU binutils development
team leader, why I couldn't build kaffe 0.9.[012] on IRIX 5.2 if
threads.s is compiled with GNU as.  Compiling it with IRIX's built-in
as would do fine.

We couldn't decide whose fault it was exactly, but it is possible that
it is a bug in IRIX's ld, since the same object file, produced by GNU
as 2.8.1, is accepted by IRIX 5.3's ld.

Anyway, Ian has suggested the attached patch for threads.s.  It does
not affect the outcome at all, but it provides more accurate
information about the assembly code.

Regards,

-- 
Alexandre Oliva
mailto:oliva@dcc.unicamp.br mailto:aoliva@acm.org
http://www.dcc.unicamp.br/~oliva
Universidade Estadual de Campinas, SP, Brasil

--Multipart_Mon_Oct_20_20:10:09_1997-1
Content-Type: application/octet-stream; type=patch
Content-Disposition: attachment; filename="mipskaffe.diff"
Content-Transfer-Encoding: 7bit

--- config/mips/threads.s~	Fri Oct 10 20:38:39 1997
+++ config/mips/threads.s	Fri Oct 10 20:38:16 1997
@@ -11,7 +11,7 @@
  *
  */
 
-	.globl	_threadswitch
+	.globl	_threadswitch	.text
 _threadswitch:	
 	subu	$sp,$sp,32*4
 	sw	$2,  2*4($sp)
@@ -77,7 +77,7 @@
 	addu	$sp,$sp,32*4	
 	j	$31
 
-	.globl	_threadsp	
+	.globl	_threadsp	.text
 _threadsp:
 	move	$2,$sp
 	j	$31

--Multipart_Mon_Oct_20_20:10:09_1997-1--
From kaffe@kaffe.org  Tue Oct 21 03:33:13 1997
From: kaffe@kaffe.org (Michael Gesundheit)
Date: Mon, 20 Oct 1997 19:33:13 -0700 (PDT)
Subject: Kaffe bug on PC
Message-ID: <19971021023313.12883.rocketmail@send1.rocketmail.com>


Hi There,

I was able to build kaffe under VC++ 5.0. It does not
work yet but looks promissing. I have the following
problem:

In initialization time while executing initialization
of system class the vm gets into nested calls to
itself situation. In file kaffe.def:

define_insn(INVOKESTATIC)
{
......

#if defined(TRANSLATOR) && defined(HAVE_TRAMPOLINE))
  
call_indirect_const(&METHOD_NATIVECODE(method_method());
#else
   call(tmp);
#endif

........
}

In my configuration the call(tmp) is what compiled
and, if you look at this macro it is, basicly, calling
virtualMachine() which again gets back to INVOKESTATIC
case and like that - forever until the program runs
out of stack.

Any clue????

Thanks,

--Michael




_____________________________________________________________________
Sent by RocketMail. Get your free e-mail at http://www.rocketmail.com

From kaffe@kaffe.org  Tue Oct 21 18:18:27 1997
From: kaffe@kaffe.org (George Chung)
Date: Tue, 21 Oct 1997 10:18:27 -0700
Subject: Native threads and native methods on Solaris
Message-ID: <01BCDE0A.AC65E750.gchung@openhorizon.com>

Hi,
I'm new to this mailing list and couldn't really find anything in the 
archives, so here goes...

1) From looking at config/sparc/solaris2/*.c, it appears that the 
solaris/sparc implementation of Kaffe is based on a user level threads 
package. Is this correct?

2) If native methods are supported, either through JNI or through the old 
1.02 technique, then the native method library must be linked with this 
Kaffe user level threads package. Is this correct? Do you provide jacket 
routines for the synchronous system calls such as read, write, etc?

Regards,
______________________________________________________________________
George Chung
Open Horizon, Inc.
601 Gateway Blvd.
Suite 800
South San Francisco, CA 94080
t: (650) 869-2220
f: (650) 869-2201
e: gchung@openhorizon.com
w: http://www.openhorizon.com

Find out about Ambrosia(tm), the SECURE Publish/Subscribe Event
Management System for the Internet (and 100% Pure Java Certified),
at http://www.openhorizon.com.

From kaffe@kaffe.org  Tue Oct 21 19:10:33 1997
From: kaffe@kaffe.org (David Waite)
Date: Tue, 21 Oct 1997 14:10:33 -0400
Subject: ICQ for Java success?
Message-ID: <199710211839.OAA29478@m23.rickards.leon.k12.fl.us>

Has anyone had success in getting ICQ (http://www.mirabilis.com) to run
under kaffe? I've tried to get it started several times, there is a problem
with the released binaries in the linux jdk that prevents me from using them
with my system. I haven't tried kaffe yet though

ICQ is an internet pager concept: you add users to you rlist, then the
server tells you when they are on, and lets you send messages/ files and
chat with them.

-David Waite

From kaffe@kaffe.org  Wed Oct 22 08:28:50 1997
From: kaffe@kaffe.org (Nathan Gelbard)
Date: Wed, 22 Oct 1997 00:28:50 -0700
Subject: NetBSD-m68k-0.9.1 compile problem
Message-ID: <3.0.32.19971022002850.007dee70@engr.orst.edu>

I'm running NetBSD-m68k-hp300 1.2E

gcc -g -O  -I. -I. -I./../../config -I../../config -I../../include
-I./../../inc
lude -DINTERPRETER -I./intrp -DKVER=\"0.91\" -c -fpic ./intrp/machine.c
./intrp/machine.c: In function `virtualMachine':
./intrp/machine.c:141: structure has no member named `exception_table_len'
./intrp/machine.c:152: structure has no member named `ins'
./intrp/machine.c:178: structure has no member named `code'
./intrp/machine.c:182: structure has no member named `codelen'

I just grabbed the src tonight.

What to do?

	nate



[ Nathan Gelbard              ][ gelbard@engr.orst.edu           ] 
[ Oregon State University     ][ http://straylight.hurrah.com    ]
[ Student of Computer Science ][ Area Support Computer Assistant ]
From kaffe@kaffe.org  Wed Oct 22 19:03:49 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Wed, 22 Oct 1997 12:03:49 -0600 (MDT)
Subject: intrp/machine.c bug
Message-ID: <199710221803.MAA04478@lal.cs.utah.edu>

Hi,

I was encountering wacky behavior in my hacked copy of Kaffe when
running lots of threads in some simple tests.  (I couldn't get the
release version to duplicate the problem, but I think its still a
problem in 0.9.2.)  Basically, a bunch of stack variables were getting
trashed in intrp/machine.c::virtualMachine() (the interpreter loop.)
I ran 'gcc -O2 -Wall' on it, and gcc tells me that a bunch of
variables might get trashed by longjmp()---exactly the variables that
were getting trashed.  There is a single setjmp() in virtualMachine().
I got rid of the warnings (and the problem!) by putting the setjmp in
a separate function (I read that the problems with setjmp/longjmp are
limited to the active stack frame).  Attached is a patch for
intrp/machine.c.  Anyone with more experience with setjmp/longjmp care
to explain what was going on?  ;)

-Pat

--- /home/grad/tullmann/java/kaffe-0.9.2/kaffe/kaffevm/intrp/machine.c Sat Oct  4 04:26:57 1997
+++ /home/grad/tullmann/tmp/machine.c   Wed Oct 22 11:59:46 1997
@@ -68,6 +68,12 @@
 /* Misc stuff */
 Hjava_lang_Object* exceptionObject;
 
+static int
+exceptionSetjmp(vmException *mjbuf)
+{
+       return setjmp(mjbuf->jbuf);
+}
+
 void
 virtualMachine(methods* meth, slots* arg, slots* retval)
 {
@@ -140,7 +146,7 @@
                TCTX(currentThread)->exceptPtr = &mjbuf;
        }
        if (meth->exception_table != 0 || (methaccflags & ACC_SYNCHRONISED)) {
-               if (setjmp(mjbuf.jbuf) != 0) {
+               if (exceptionSetjmp(&mjbuf) != 0) {
                        assert(TCTX(currentThread)->exceptPtr == &mjbuf);
                        npc = mjbuf.pc;
                        sp = &lcl[meth->localsz + meth->stacksz - 1];

From kaffe@kaffe.org  Wed Oct 22 19:44:21 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Wed, 22 Oct 1997 12:44:21 -0600 (MDT)
Subject: bug in intrp/machine.c
Message-ID: <199710221844.MAA04631@lal.cs.utah.edu>

Doh!

I didn't test that solution quite as well as I should have!

setjmp/longjmp specify that you cannot return from a function that you
call setjmp in and expect it to work.  (So one problem goes away, but
Kaffe's exception handling ability is trashed.)  The standard solution
is to make all the questionable stack variables volatile...  I'll see
what that does...

-Pat



From kaffe@kaffe.org  Wed Oct 22 20:32:19 1997
From: kaffe@kaffe.org (Amitabh Arora)
Date: Wed, 22 Oct 1997 12:32:19 -0700
Subject: Binaries for Sun 4.1.3 and HP-UX 9.0 OS's
Message-ID: <01BCDEE6.8A8C0DD0@amitabh@aspec.com>

As the title says.  Does any body have the binaries for the above two
platforms.  It will save me a lot of work and I can spend more time writing
Java applications.  My compilations have failed in the past and I am unable
to devote time to fix the compiles.

Please please help.

Thanks. -amitabh arora
From kaffe@kaffe.org  Wed Oct 22 16:06:31 1997
From: kaffe@kaffe.org (Nikolai Vladychevski)
Date: Wed, 22 Oct 97 15:06:31 -0000
Subject: NetBSD-m68k-0.9.1 compile problem
Message-ID: <199710221957.UAA18996@home.virtual-pc.com>

>Subject:     NetBSD-m68k-0.9.1 compile problem
>Sent:        10/22/97 7:27
>Received:    10/22/97 9:33
>From:        Nathan Gelbard, gelbard@engr.orst.edu
>Reply-To:    kaffe@kaffe.org
>To:          kaffe@sarc.city.ac.uk
>
>I'm running NetBSD-m68k-hp300 1.2E
>
>gcc -g -O  -I. -I. -I./../../config -I../../config -I../../include
>-I./../../inc
>lude -DINTERPRETER -I./intrp -DKVER=\"0.91\" -c -fpic ./intrp/machine.c
>./intrp/machine.c: In function `virtualMachine':
>./intrp/machine.c:141: structure has no member named `exception_table_len'
>./intrp/machine.c:152: structure has no member named `ins'
>./intrp/machine.c:178: structure has no member named `code'
>./intrp/machine.c:182: structure has no member named `codelen'
>
>I just grabbed the src tonight.
>
>What to do?
>
>	nate
this question was answered in this list already, but anyway, i couldnt run
kaffe 9.1 on my Mac, due to problem with symbols of some libraries, this 
was
also discussed in this list, but when i compiled it, and ran, it told me 
about some
symbols problem. I didnt insist, but if you make it work, let me know i 
will fight again.

niko

From kaffe@kaffe.org  Sat Oct 25 02:08:49 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Sat, 25 Oct 1997 02:08:49 +0100
Subject: A Beautiful Painting!
Message-ID: <199710250108.CAA25266@cripplecock.sarc.city.ac.uk>

Dear Internet User,

Are you one of those people who love beautiful things around you?

Why not invest in an original oil painting?

For more information reply to:

ART@mfcnet.net

Sincerely Yours,

Phil & Jackie Clark

If you would like to be removed from our list, please send an
Email to remove@mfcnet.net with "remove" in the subject line.




From kaffe@kaffe.org  Sat Oct 25 23:24:13 1997
From: kaffe@kaffe.org (Matthias Hopf (@Home))
Date: Sat, 25 Oct 1997 22:24:13 MET1
Subject: NetBSD-m68k-0.9.1 compile problem
In-Reply-To: <199710221957.UAA18996@home.virtual-pc.com>
Message-ID: <yam7237.651.123577448@faui01.informatik.uni-erlangen.de>

Hi Nikolai!

> >I'm running NetBSD-m68k-hp300 1.2E
> >
> this question was answered in this list already, but anyway, i couldnt run
> kaffe 9.1 on my Mac, due to problem with symbols of some libraries, this 
> was
> also discussed in this list, but when i compiled it, and ran, it told me 
> about some
> symbols problem. I didnt insist, but if you make it work, let me know i 
> will fight again.

Sorry, I don't have a NextStep computer and thus I am not able to help you.
I only wrote the original m68k-assembler macros (for AmigaOS) which were
used for the NextStep port. Best ask at the mailing list kaffe@kaffe.org
(majordomo) for help.

Don't know why I stand in NOTES/HELPERS as NeXT porter :-]

Sorry

Matthias

-- 
    //             |   Matthias Hopf - "Hoeppel"   |    _      __
\\ //    Amiga     |  student of computer science  |  _|cience |-iction
 \X/ by conviction |      in Erlangen/Germany      | by belief in Future

 EMail:  mshopf@informatik.uni-erlangen.de
 Aminet: ftpamiga@epix.rrze.uni-erlangen.de
 WWW:    http://wwwcip.informatik.uni-erlangen.de/user/mshopf/


From kaffe@kaffe.org  Sun Oct 26 01:53:06 1997
From: kaffe@kaffe.org (Duane C. Cato)
Date: Sat, 25 Oct 1997 20:53:06 -0400
Subject: A Beautiful Painting!
Message-ID: <3.0.32.19971025205240.012980d4@mail.geocities.com>

At 02:08 AM 10/25/97 +0100, you wrote:
>Dear Internet User,
>
>Are you one of those people who love beautiful things around you?
>
>Why not invest in an original oil painting?
>
>For more information reply to:
>
>ART@mfcnet.net
>
>Sincerely Yours,
>
>Phil & Jackie Clark
>
>If you would like to be removed from our list, please send an
>Email to remove@mfcnet.net with "remove" in the subject line.

Look - PHIL!
Don't Spam, it is annoying, and simply uses up our listserver space. If we
want your stuff, we can and WILL find you.
Remove this listserver address from your list!

From kaffe@kaffe.org  Sun Oct 26 14:26:23 1997
From: kaffe@kaffe.org (Matthias Hopf (@Home))
Date: Sun, 26 Oct 1997 14:26:23 MET1
Subject: NetBSD-m68k-0.9.1 compile problem
In-Reply-To: <yam7237.651.123577448@faui01.informatik.uni-erlangen.de>
Message-ID: <yam7238.1178.124108720@faui01.informatik.uni-erlangen.de>

Hihi!

Matthias wrote:

> Sorry, I don't have a NextStep computer and thus I am not able to help you.
> I only wrote the original m68k-assembler macros (for AmigaOS) which were
> used for the NextStep port. Best ask at the mailing list kaffe@kaffe.org
> (majordomo) for help.

Grrr, should take a look at the header next time I'm replying to a
message... :-]

> Don't know why I stand in NOTES/HELPERS as NeXT porter :-]

This is still correct, though. I thought I should send a message to Tim
after my final exams (tuesday), but then, I can do it immedeately...

Tim, could you please remove me as NeXT porter in NOTES/HELPERS? I'm
repeatedly getting questions about the NeXT port, and I can't help those
folks. I don't know, who's done the NeXT port, either...

I'm still responsible for the Amiga port, though.

Cheers

Matthias

-- 
    //             |   Matthias Hopf - "Hoeppel"   |    _      __
\\ //    Amiga     |  student of computer science  |  _|cience |-iction
 \X/ by conviction |      in Erlangen/Germany      | by belief in Future

 EMail:  mshopf@informatik.uni-erlangen.de
 Aminet: ftpamiga@epix.rrze.uni-erlangen.de
 WWW:    http://wwwcip.informatik.uni-erlangen.de/user/mshopf/


From kaffe@kaffe.org  Sun Oct 26 17:55:15 1997
From: kaffe@kaffe.org (Martin Lanzen)
Date: Sun, 26 Oct 1997 18:55:15 +0100 (MET)
Subject: remove
In-Reply-To: <199710250108.CAA25266@cripplecock.sarc.city.ac.uk>
Message-ID: <199710261755.SAA14533@puh>

On 25 Okt , wwp@mfcnet.net wrote:
> Dear Internet User,
> 
> Are you one of those people who love beautiful things around you?
> 
> Why not invest in an original oil painting?
> 
> For more information reply to:
> 
> ART@mfcnet.net
> 
> Sincerely Yours,
> 
> Phil & Jackie Clark
> 
> If you would like to be removed from our list, please send an
> Email to remove@mfcnet.net with "remove" in the subject line.
> 
> 
> 
> 

-- 
                                                           
/Martin                                          \ \ | / /
                                                  ( @ @ )
--------------------------------------------o00o----()----o00o-----
| Martin Lanzen                                                      |
| Phone: 0303-19384,031-7036178,0707-414757                          |
| OBS --> New adress: Åsgatan 16, 442 30 Kungälv  <-- OBS            |
| Email: e4lanzen@etek.chalmers.se              .oooO                |
| http://www.etek.chalmers.se/~e4lanzen/         (   )   Oooo.       |
--------------------------------------------------\ (----(   )--------
                                                   \_)    ) /
                                                         (_/


From kaffe@kaffe.org  Sun Oct 26 20:04:16 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Mon, 27 Oct 1997 07:04:16 +1100
Subject: OtS Music Mag & Classifieds
Message-ID: <199710262004.HAA00917@Hi-Fi.infinet.net.au.>

Dear Member,

Edition 871 of the music Mag is on line.

Classifieds are now working from the site with over 50

classifications at http://www.onthestreet.com

There is no charge for using our classified service.

You can advertise anything including Gigs.

If you want to be deleted from our member service send an email 
with "remove" in the subject.

Thanks,

On The Street Magazine
Phone 02 9211 1122
Australia
From kaffe@kaffe.org  Mon Oct 27 06:59:23 1997
From: kaffe@kaffe.org (Danilov Nikita)
Date: Mon, 27 Oct 1997 09:59:23 +0300
Subject: remove
References: <199710262004.HAA00917@Hi-Fi.infinet.net.au.>
Message-ID: <34543BCB.38DA7CC8@server.ru>

This is a multi-part message in MIME format.
--------------CF7FE7283E109198A5AD69FD
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

ots@infinet.net.au wrote:
> 
> Dear Member,
> 
> Edition 871 of the music Mag is on line.
> 
> Classifieds are now working from the site with over 50
> 
> classifications at http://www.onthestreet.com
> 
> There is no charge for using our classified service.
> 
> You can advertise anything including Gigs.
> 
> If you want to be deleted from our member service send an email
> with "remove" in the subject.
> 
> Thanks,
> 
> On The Street Magazine
> Phone 02 9211 1122
> Australia

-- 
Mail me: god@server.ru                  Phone  me:  +7(095)4348205
Web  me: http://finn.server.ru/~god     Fax    me:  +7(095)4349267
         http://www.server.ru/~god
Of the first was he to bare arms and a name: 
Wassaily Booslaeugh of Riesengeborg. // J.Joyce, Finnegans Wake
--------------CF7FE7283E109198A5AD69FD
Content-Type: text/x-vcard; charset=us-ascii; name="vcard.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Nikita Danilov
Content-Disposition: attachment; filename="vcard.vcf"

begin:          vcard
fn:             Nikita Danilov
n:              Danilov;Nikita
org:            Server Ltd.
adr:            Server Ltd.,;;78 Vernardsky pr.;Moscow;Moscow region;117454;Russia
email;internet: god@server.ru
title:          Expert
tel;work:       +7(095)4348205
tel;fax:        +7(095)4348205
x-mozilla-cpt:  ;0
x-mozilla-html: TRUE
version:        2.1
end:            vcard


--------------CF7FE7283E109198A5AD69FD--

From kaffe@kaffe.org  Mon Oct 27 09:23:08 1997
From: kaffe@kaffe.org (Danilov Nikita)
Date: Mon, 27 Oct 1997 12:23:08 +0300
Subject: remove
References: <199710262004.HAA00917@Hi-Fi.infinet.net.au.> <34543BCB.38DA7CC8@server.ru>
Message-ID: <34545D7C.187335A4@server.ru>

This is a multi-part message in MIME format.
--------------AA55EB946375B9432B39F822
Content-Type: text/plain; charset=koi8-r
Content-Transfer-Encoding: 7bit

Sorry, next time I shall see to the headers thoroughly

Danilov Nikita wrote:
> 
> ots@infinet.net.au wrote:
> >
> > Dear Member,
...

Nikita.
-- 
Mail me: god@server.ru                  Phone  me:  +7(095)4348205
Web  me: http://finn.server.ru/~god     Fax    me:  +7(095)4349267
         http://www.server.ru/~god
Of the first was he to bare arms and a name: 
Wassaily Booslaeugh of Riesengeborg. // J.Joyce, Finnegans Wake
--------------AA55EB946375B9432B39F822
Content-Type: text/x-vcard; charset=koi8-r; name="vcard.vcf"
Content-Transfer-Encoding: 7bit
Content-Description: Card for Nikita Danilov
Content-Disposition: attachment; filename="vcard.vcf"

begin:          vcard
fn:             Nikita Danilov
n:              Danilov;Nikita
org:            Server Ltd.
adr:            Server Ltd.,;;78 Vernardsky pr.;Moscow;Moscow region;117454;Russia
email;internet: god@server.ru
title:          Expert
tel;work:       +7(095)4348205
tel;fax:        +7(095)4348205
x-mozilla-cpt:  ;0
x-mozilla-html: TRUE
version:        2.1
end:            vcard


--------------AA55EB946375B9432B39F822--

From kaffe@kaffe.org  Mon Oct 27 09:02:05 1997
From: kaffe@kaffe.org (Umaire benoit)
Date: Mon, 27 Oct 1997 10:02:05 +0100
Subject: remove
Message-ID: <199710270902.KAA00639@sun46.butt1>

remove

From kaffe@kaffe.org  Mon Oct 27 11:19:28 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Mon, 27 Oct 1997 22:19:28 +1100 (EST)
Subject: java.util.Date not compatible b/w kaffe-0.8.4 and 0.9.1
Message-ID: <199710271119.WAA17776@zen.quick.com.au>

It looks very much like there is something very different b/w kaffe
versions wrt Date native methods.

I have an Strftime() class which provides decent formatting for times
etc.  It currently needs to be compiled with jdk1.0.2 (kaffe-0.8.4)
and when run with that VM produces correct results:

sjg:771$ JDK=1.0.2 java au.com.quick.util.Strftime %C "%Y%m%d %T"       
default: Mon Oct 27 22:16:16  GMT-1100 1997
%C: Mon Oct 27 22:16:16  GMT-1100 1997
%Y%m%d %T: 19971027 22:16:16
sjg:772$ date
Mon Oct 27 22:16:17 EST 1997

If I run it with kaffe-0.9.1 though I get:

sjg:773$ java au.com.quick.util.Strftime %C "%Y%m%d %T"           
default: Mon Oct 27 11:17:06  GMT 1997
%C: Mon Oct 27 11:17:06  GMT 1997
%Y%m%d %T: 19971027 11:17:06
sjg:774$ date
Mon Oct 27 22:17:07 EST 1997

If JDK is not set, my java_wrapper uses the greatest installed jdk
(1.1.4 here)

I would guess that the TimeZone() stuff does not work in 0.9.1
[sorry havn't solved the core dumps in 0.9.2 so can't test it].

--sjg

From kaffe@kaffe.org  Mon Oct 27 14:25:30 1997
From: kaffe@kaffe.org (Tim Wilkinson)
Date: Mon, 27 Oct 1997 14:25:30 +0000
Subject: java.util.Date not compatible b/w kaffe-0.8.4 and 0.9.1
References: <199710271119.WAA17776@zen.quick.com.au>
Message-ID: <3454A45A.167EB0E7@tjwassoc.co.uk>

Hi,

In JDK 1.0.2 java.util.Date used a number of native methods (Kaffe
0.8.x).  But in JDK 1.1.x java.util.Date is completely written in Java
and has not Kaffe bits at all (kaffe 0.9.x).

Tim
--
  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
  London, UK.                           Email:   tim@tjwassoc.co.uk

From kaffe@kaffe.org  Mon Oct 27 19:18:20 1997
From: kaffe@kaffe.org (Mark Hamzy)
Date: Mon, 27 Oct 1997 14:18:20 -0500 (EST)
Subject: remove
In-Reply-To: <199710250108.CAA25266@cripplecock.sarc.city.ac.uk> from "wwp@mfcnet.net" at Oct 25, 97 02:08:49 am
Message-ID: <199710271918.OAA12838@grumpy.magg.net>

remove

> 
> Dear Internet User,
> 
> Are you one of those people who love beautiful things around you?
> 
> Why not invest in an original oil painting?
> 
> For more information reply to:
> 
> ART@mfcnet.net
> 
> Sincerely Yours,
> 
> Phil & Jackie Clark
> 
> If you would like to be removed from our list, please send an
> Email to remove@mfcnet.net with "remove" in the subject line.
> 
> 
> 
> 


-- 
Anyone failing to display the mandatory black velvet Methodist style painting
   of President Quayle on their living room wall... WILL BE SHOT.
Defenestration n.: [ModL defenestratio < L de + fenestra, window]
   a throwing out through a window

From kaffe@kaffe.org  Mon Oct 27 18:25:58 1997
From: kaffe@kaffe.org (Tim Wilkinson)
Date: Mon, 27 Oct 1997 18:25:58 +0000
Subject: [TEST] New majordomo
Message-ID: <3454DCB6.41C67EA6@tjwassoc.co.uk>

This is a test message since we've just moved the majorodomo to w3.org.

Tim
--
  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
  London, UK.                           Email:   tim@tjwassoc.co.uk

From kaffe@kaffe.org  Mon Oct 27 21:41:24 1997
From: kaffe@kaffe.org (Mary Valente)
Date: Mon, 27 Oct 1997 13:41:24 -0800
Subject: remove
References: <199710270902.KAA00639@sun46.butt1>
Message-ID: <34550A84.D20097AE@nbn.com>

(U) maire benoit wrote:
> 
> remove
O got your email. I'm just trying to get off anything regarding this
kaffe stuff. I got on by mistake. I have no idea what it is. THANKS>

From kaffe@kaffe.org  Mon Oct 27 22:00:33 1997
From: kaffe@kaffe.org (David Wozmak)
Date: Mon, 27 Oct 1997 18:00:33 -0400
Subject: remove
Message-ID: <v01530501b07abdbfd397@[192.68.0.2]>

Hey, folks!

I'd like to poke my pointy little head in here and remind everyone that,
since this is a listserv, posting a "remove" message to this list to
respond to the spam that came in here, is not going to accomplish anything.
Nor will sending a "remove" message to the spammer with your email
address.  The spammer has 'kaffe@kaffe.org' listed as the recipient, and
doesn't know a gosh-derned thing about all of you individually.

and if any of the seven or eight of you  that DID reply to the list with
remove messages, wants to be removed from the listserv, you should follow
these directions:

If you ever want to remove yourself from this mailing list,
send the following command in email to
<kaffe-request@soi.city.ac.uk>:

    unsubscribe

Or you can send mail to <Majordomo@soi.city.ac.uk> with the following
command in the body of your email message:

    unsubscribe kaffe your_name@your_provider.



>remove
>
>>
>> Dear Internet User,
>>
>> Are you one of those people who love beautiful things around you?
>>
>> Why not invest in an original oil painting?
>>
>> For more information reply to:
>>


From kaffe@kaffe.org  Mon Oct 27 22:43:54 1997
From: kaffe@kaffe.org (Bruce K. Partlow)
Date: Mon, 27 Oct 1997 14:43:54 -0800
Subject: remove
Message-ID: <01BCE2E6.C019C100@brucep.datapac.com>

remove

From kaffe@kaffe.org  Sun Oct 26 15:14:56 1997
From: kaffe@kaffe.org (Mikael Ståldal)
Date: Sun, 26 Oct 1997 16:14:56 +0100
Subject: Kaffe 0.9.2: error in kaffe/kaffe/Makefile
Message-ID: <w51U0EpfhGyX092yn@ingen.reklam.nada.kth.se>

When the kaffe executable is build, the LDFLAGS variable isn't used. Why
not? It's used when building kaffeh.

And in the same Makefile, make clean tries to remove $(LIB) which isn't
defined, not good.

Here is a patch:

*** E:\unix\kaffe-0.9.2\kaffe\kaffe\Makefile.in	Thu Jun 12 06:11:36 1997
--- Makefile.in	Sun Oct 26 20:03:10 1997
***************
*** 20,25 ****
--- 20,26 ----
  INCLUDES=	-I. -I$(srcdir)/../kaffevm -I../../config -I$(srcdir)/../../config -I../../include -I$(srcdir)/../../include
  CFLAGS=		@CFLAGS@
  ALL_CFLAGS=	$(CFLAGS) @KAFFEFLAGS@ $(INCLUDES)
+ LDFLAGS=	@LDFLAGS@
  LIBS=		@NOSHAREDLIBS@
  OBJEXT=		@OBJEXT@
  EXEEXT=		@EXEEXT@
***************
*** 29,41 ****
  all:		kaffe$(EXEEXT)
  
  kaffe$(EXEEXT):	$(OBJECT)
! 		$(CC) $(ALL_CFLAGS) -o kaffe$(EXEEXT) $(OBJECT) $(LIBS)
  
  .c$(OBJEXT):
  		$(CC) $(ALL_CFLAGS) -c $<
  
  clean:
! 		rm -f $(LIB) kaffe$(EXEEXT) kaffe.core $(OBJECT)
  
  distclean:	clean
  		rm -f Makefile
--- 30,42 ----
  all:		kaffe$(EXEEXT)
  
  kaffe$(EXEEXT):	$(OBJECT)
! 		$(CC) $(LDFLAGS) -o kaffe$(EXEEXT) $(OBJECT) $(LIBS)
  
  .c$(OBJEXT):
  		$(CC) $(ALL_CFLAGS) -c $<
  
  clean:
! 		rm -f kaffe$(EXEEXT) kaffe.core $(OBJECT)
  
  distclean:	clean
  		rm -f Makefile

-- 
/****************************************************************\
* You have just read a message from Mikael Ståldal.              *
*                                                                *
* Remove "ingen.reklam." from the address before mail replying.  *
\****************************************************************/

From kaffe@kaffe.org  Sun Oct 26 18:52:15 1997
From: kaffe@kaffe.org (Mikael Ståldal)
Date: Sun, 26 Oct 1997 19:52:15 +0100
Subject: Kaffe 0.9.2 on OS/2
Message-ID: <fF5U0EpfhK4W092yn@ingen.reklam.nada.kth.se>

Here are some patches for OS/2. To get the config/i386/os2 direcrory, just
copy config/i386/win32 and rename the cygnus subdirectory to emx.

With this patches and EMX 0.9c its possible to build Kaffe on OS/2. It
doesn't work though, it crashes as soon as I try to use it. I tried both
interpreter and JIT.

Only in kaffe/config/i386: os2
diff -cbr kaffe-0.9.2/config.guess kaffe/config.guess
*** kaffe-0.9.2/config.guess	Thu Jun 26 08:20:10 1997
--- kaffe/config.guess	Sun Oct 26 18:47:58 1997
***************
*** 442,447 ****
--- 442,450 ----
      i*86:CYGWIN32/*:*:*)
  	echo i386-unknown-win32/cygnus
  	exit 0 ;;
+     i*86:OS/2:*:*)
+ 	echo i386-pc-os2/emx
+ 	exit 0 ;;
  
  esac
  
diff -cbr kaffe-0.9.2/config.sub kaffe/config.sub
*** kaffe-0.9.2/config.sub	Thu Jun 26 08:20:12 1997
--- kaffe/config.sub	Sun Oct 26 18:45:38 1997
***************
*** 697,702 ****
--- 697,704 ----
  	-xenix)
  		os=-xenix
  		;;
+         -os2*)
+ 		;;
  	-none)
  		;;
  	*)
diff -cbr kaffe-0.9.2/configure.in kaffe/configure.in
*** kaffe-0.9.2/configure.in	Sat Oct  4 03:26:52 1997
--- kaffe/configure.in	Tue Oct 21 22:42:48 1997
***************
*** 141,146 ****
--- 141,155 ----
  	LN_S="cp"
  	;;
  
+ i[3456]86-os2/emx)
+ 	host_cpu=i386
+ 	host_os=os2/emx
+ 	EXEEXT=.exe
+ 	SCREXT=.cmd
+ 	dynamic_libraries=no
+ 	LN_S="cp"
+ 	;;
+ 
  i[3456]86-dgux*)
  	host_cpu=i386
  	host_os=dgux
***************
*** 457,462 ****
--- 466,473 ----
  case $host_os in
  win32*)
  	SYSTEM=win32 ;;
+ os2*)
+ 	SYSTEM=os2 ;;
  *)
  	SYSTEM=unix ;;
  esac
***************
*** 1060,1065 ****
--- 1071,1085 ----
  	fi
  	echo set KAFFEHOME=$datadir/kaffe >> ENVIRONMENT.BAT
  	echo set PATH=%PATH%\;$bindir >> ENVIRONMENT.BAT
+ elif test "$SYSTEM" = "os2" ; then
+ 	echo @echo off > ENVIRONMENT.CMD
+ 	if test "$awt_toolkit" != "none" ; then
+ 		echo set CLASSPATH=.\;$datadir/kaffe/classes.zip\;$datadir/kaffe/$awt_toolkit >> ENVIRONMENT.CMD
+ 	else
+ 		echo set CLASSPATH=.\;$datadir/kaffe/classes.zip >> ENVIRONMENT.CMD
+ 	fi
+ 	echo set KAFFEHOME=$datadir/kaffe >> ENVIRONMENT.CMD
+ 	echo set PATH=%PATH%\;$bindir >> ENVIRONMENT.CMD
  fi
  
  dnl -------------------------------------------------------------------------
diff -cbr kaffe-0.9.2/include/system.h kaffe/include/system.h
*** kaffe-0.9.2/include/system.h	Mon May  5 07:25:32 1997
--- kaffe/include/system.h	Sun Oct 26 15:05:22 1997
***************
*** 22,28 ****
  #define	file_seperator		"/"
  #define	path_seperator		":"
  #define	line_seperator		"\n"
! #elif defined(__WIN32__) || defined(__OS2__)
  #define	file_seperator		"\\"
  #define	path_seperator		";"
  #define	line_seperator		"\r\n"
--- 22,28 ----
  #define	file_seperator		"/"
  #define	path_seperator		":"
  #define	line_seperator		"\n"
! #elif defined(__WIN32__) || defined(__OS2__) || defined(__EMX__)
  #define	file_seperator		"\\"
  #define	path_seperator		";"
  #define	line_seperator		"\r\n"
diff -cbr kaffe-0.9.2/kaffe/kaffevm/paths.h kaffe/kaffe/kaffevm/paths.h
*** kaffe-0.9.2/kaffe/kaffevm/paths.h	Wed Apr 16 01:10:56 1997
--- kaffe/kaffe/kaffevm/paths.h	Sun Oct 26 15:07:30 1997
***************
*** 18,24 ****
  #define	MAXPATHLEN	1024
  #endif
  
! #if defined(__WIN32__) || defined(__OS2__)
  #define	PATHSEP		';'
  #define	DIRSEP		"\\"
  #elif defined(__amigaos__)
--- 18,24 ----
  #define	MAXPATHLEN	1024
  #endif
  
! #if defined(__WIN32__) || defined(__OS2__) || defined(__EMX__)
  #define	PATHSEP		';'
  #define	DIRSEP		"\\"
  #elif defined(__amigaos__)

-- 
/****************************************************************\
* You have just read a message from Mikael Ståldal.              *
*                                                                *
* Remove "ingen.reklam." from the address before mail replying.  *
\****************************************************************/

From kaffe@kaffe.org  Mon Oct 27 21:45:54 1997
From: kaffe@kaffe.org (Mikael Ståldal)
Date: Mon, 27 Oct 1997 22:45:54 +0100
Subject: Problem with Makefiles: SHELL missing
Message-ID: <SuQV0Epfha0Y092yn@nada.kth.se>

The Makefiles make use of UNIX shell features, but they doesn't contain any
SHELL caluse. They should have a

SHELL = /bin/sh

line.

-- 
/****************************************************************\
* You have just read a message from Mikael Ståldal.              *
*                                                                *
* Remove "ingen.reklam." from the address before mail replying.  *
\****************************************************************/

From kaffe@kaffe.org  Tue Oct 28 05:37:30 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Tue, 28 Oct 1997 16:37:30 +1100 (EST)
Subject: java.util.Date not compatible b/w kaffe-0.8.4 and 0.9.1
References: <199710271119.WAA17776@zen.quick.com.au> <3454A45A.167EB0E7@tjwassoc.co.uk>
Message-ID: <199710280537.QAA15984@gate.quick.com.au>

>In JDK 1.0.2 java.util.Date used a number of native methods (Kaffe
>0.8.x).  But in JDK 1.1.x java.util.Date is completely written in Java
>and has not Kaffe bits at all (kaffe 0.9.x).

Thanks. Turns out that for jdk1.1.x you need to have user.timezone
set.  Naffed if I can find any doco which says what file java looks
in for that sort of thing... by default that is.

Oh if only they had kept getenv :-)

--sjg
-- 
Simon J. Gerraty        <sjg@quick.com.au>

#include <disclaimer>   /* imagine something _very_ witty here */

From kaffe@kaffe.org  Tue Oct 28 07:56:34 1997
From: kaffe@kaffe.org (Holger Veit)
Date: Tue, 28 Oct 1997 08:56:34 +0100 (MET)
Subject: Kaffe 0.9.2 on OS/2
In-Reply-To: <fF5U0EpfhK4W092yn@ingen.reklam.nada.kth.se> from "Mikael St\eldal" at Oct 26, 97 07:52:15 pm
Message-ID: <199710280756.AA00548@simi.gmd.de>

> 
> Here are some patches for OS/2. To get the config/i386/os2 direcrory, jus=
> t
> copy config/i386/win32 and rename the cygnus subdirectory to emx.
> 
> With this patches and EMX 0.9c its possible to build Kaffe on OS/2. It
> doesn't work though, it crashes as soon as I try to use it. I tried both
> interpreter and JIT.

Congratulations: you have likely found the same problem than I quite some
time ago, however I haven't bothered to try out the interpreter. So my
initial assumption of an obscure problem in the JIT or JIT-generated code
seems to be wrong. This makes room for other speculation: kaffe emulates
threading, rather than relying on native threads or pthread packages. I
had annotated the class loader and found the system crashing in different
classes which might look like a timing problem. I haven't seen a native
threads application in the tree yet (however this was 0.9.1, maybe this has
changed in 0.9.2), so a good idea might be to replace the internal threading
stuff with native threads. You can find a snapshot of my current 0.91 tree
at ftp://set.gmd.de/pub/misc/XFree86OS2/test/kaffe.zip for reference
(508K, directory readable; note it is not worthwhile to download it unless one 
is really interested in porting - it doesn't work).

-- 
         Dr.-Ing. Holger Veit             | INTERNET: Holger.Veit"at"gmd.de
|  |   / GMD - German National Research   | Phone: (+49) 2241 14 2448
|__|  /  Center for Information Technology| Fax:   (+49) 2241 14 2242
|  | /   Schloss Birlinghoven             |  Get XFree86/OS2 Bugfix Version
|  |/    D-53754 Sankt Augustin, Germany  |    V3.3.1 from ftp.xfree86.org
         WWW: http://borneo.gmd.de/~veit/ | /pub/XFree86/3.3.1/binaries/OS2

From kaffe@kaffe.org  Tue Oct 28 15:16:07 1997
From: kaffe@kaffe.org (Robert Bates)
Date: Tue, 28 Oct 1997 09:16:07 -0600
Subject: QNX 4.x port of 0.9.2...?
Message-ID: <3.0.3.32.19971028091607.0093b7d0@zeh.com>

Has anyone had any more luck getting kaffe 0.9.2 to port to QNX?  I'm about
to go out and sink the $$$ into the QNX devel system, and would like to get
a Java VM up and running ASAP on that platform.  Seeing as how QNX has a
few more months to go on their *own* JVM... ;)

I read the article about getting 0.9.1 to build on QNX, so I'm good at
least up to that point...

Also, are there any mirrors of the gcc port to QNX referenced in the
posting?  I'm having a hard time getting a complete d/l from the France
site...

TIA,
Rob

//////////////////////////////////////////////////////////////////////////
// Robert Bates
// Infosys Development Group
// Houston, Texas  USA
//
// rbates@iname.com
// http://home.att.net/~rpbates
//////////////////////////////////////////////////////////////////////////
#ifdef _LITIGATORS_PRESENT_
#include <std_disclaimer.h>
#endif

From kaffe@kaffe.org  Tue Oct 28 17:29:49 1997
From: kaffe@kaffe.org (Robert Bates)
Date: Tue, 28 Oct 1997 11:29:49 -0600
Subject: Kaffe 0.9.2 on QNX 4.x...?
Message-ID: <3.0.3.32.19971028112949.009824d0@zeh.com>

Has anyone had any more luck getting kaffe 0.9.2 to port to QNX?  I'm about
to go out and sink the $$$ into the QNX devel system, and would like to get
a Java VM up and running ASAP on that platform.  Seeing as how QNX has a
few more months to go on their *own* JVM... ;)

I read the article about getting 0.9.1 to build on QNX, so I'm good at
least up to that point...

Also, are there any mirrors of the gcc port to QNX referenced in the
posting?  I'm having a hard time getting a complete d/l from the France
site...

TIA,
Rob

//////////////////////////////////////////////////////////////////////////
// Robert Bates
// Infosys Development Group
// Houston, Texas  USA
//
// rbates@iname.com
// http://home.att.net/~rpbates
//////////////////////////////////////////////////////////////////////////
#ifdef _LITIGATORS_PRESENT_
#include <std_disclaimer.h>
#endif

From kaffe@kaffe.org  Tue Oct 28 18:11:09 1997
From: kaffe@kaffe.org (Yuang Shuang-Luang)
Date: Wed, 29 Oct 1997 02:11:09 +0800 (CST)
Subject: Failed to locate native function: getFontPeer
Message-ID: <199710281811.CAA10185@Edward.net>


>Submitter-Id:  kaffe-user
>Originator: 	Yuang Shuang-Luang
>Organization:	
>Priority:	medium
>Category: 	
>Class:		sw-bug 
>Release:	
>Environment:

	System: FreeBSD Edward 3.0-CURRENT FreeBSD 3.0-CURRENT #0: Mon Oct 27 18:55:20 CST 1997     edward@Edward:/usr/src/sys/compile/Ed  i386
	Architecture: i386
	Machine: pentium
>Description:

	Thw X-win version I used is 3.3.1, when I try to run appletviewer,
	it doesn't work and show the following error message:

	Failed to locate native function:
	java/awt/Toolkit.getFontPeer(Ljava/lang/String;I)Ljava/awt/peer/FontPeer
	;
	java.lang.UnsatisfiedLinkError
	at java/awt/Font.<init>(119)
	at sun/applet/AppletCopyright.<init>(40)
	at sun/applet/AppletViewer.mainInit(984)
	at sun/applet/AppletViewer.main(993)

	How can I make appletviewer work? Should I set some environment 
	variable to fix it? Thanx...

From kaffe@kaffe.org  Tue Oct 28 20:28:16 1997
From: kaffe@kaffe.org (Mary Valente)
Date: Tue, 28 Oct 1997 12:28:16 -0800
Subject: please remove!
References: <199710271918.OAA12838@grumpy.magg.net>
Message-ID: <34564AE0.9FAF7F02@nbn.com>

Mark Hamzy wrote:
> 
> remove
> 
> >
> > Dear Internet User,
> >
> > Are you one of those people who love beautiful things around you?
> >
> > Why not invest in an original oil painting?
> >NO>>>>>PLEASE REMOVE!
> > For more information reply to:
> >
> > ART@mfcnet.net
> >
> > Sincerely Yours,
> >
> > Phil & Jackie Clark
> >
> > If you would like to be removed from our list, please send an
> > Email to remove@mfcnet.net with "remove" in the subject line.
> >
> >
> >
> >
> 
> --
> Anyone failing to display the mandatory black velvet Methodist style painting
>    of President Quayle on their living room wall... WILL BE SHOT.
> Defenestration n.: [ModL defenestratio < L de + fenestra, window]
>    a throwing out through a window

From kaffe@kaffe.org  Wed Oct 29 16:09:16 1997
From: kaffe@kaffe.org (Daniel Veillard)
Date: Wed, 29 Oct 1997 11:09:16 -0500
Subject: Kaffe bug on PC
Message-ID: <199710291609.LAA23851@rufus.w3.org>


>I was able to build kaffe under VC++ 5.0. It does not
>work yet but looks promissing. I have the following
>problem:
>
>In initialization time while executing initialization
>of system class the vm gets into nested calls to
>itself situation. In file kaffe.def:
>
>define_insn(INVOKESTATIC)
>{
>......
>
>#if defined(TRANSLATOR) && defined(HAVE_TRAMPOLINE))
>  
>call_indirect_const(&METHOD_NATIVECODE(method_method());
>#else
>   call(tmp);
>#endif
>
>........
>}
>
>In my configuration the call(tmp) is what compiled
>and, if you look at this macro it is, basicly, calling
>virtualMachine() which again gets back to INVOKESTATIC
>case and like that - forever until the program runs
>out of stack.
>
>Any clue????

   Not really, did you try to compile the interpreter or the JIT.
BTW I'm very interested in a version of Kaffe in the VC++ environment.
Did you had to patch a lot of things ?

Daniel


From kaffe@kaffe.org  Wed Oct 29 17:30:13 1997
From: kaffe@kaffe.org (Michael Gesundheit)
Date: Wed, 29 Oct 1997 09:30:13 -0800 (PST)
Subject: Kaffe bug on PC
Message-ID: <19971029173013.23424.rocketmail@web1.rocketmail.com>


I build kaffe for interpreter. Not too many patches.
It is not running yet. My current problem has to to
with the slots union.

Luckyly I have a Solaris Ultra next to my PC. I run
gdb on Solaris and VC++ on the PC.

The problem has to do with HelloWorldApp.class. The 4th
byte code is INVOKEVIRTUAL. On the PC one of the macros
produce a null pointer which is basicly the 3rd stack
entry. It should, and is on Solaris, a pointer but
NULL on the PC. I didn't pinpoint the problem yet but
it has to do with the difference between the compilers
and the way they handle unions. The size of the slots
union is 8 bytes. I see the pointer in the upper 4
bytes in both the PC and Solaris. gcc takes the pointer
correctly. VC++ takes the lower 4 bytes which are NULL.
I'll address this problem today and broadcast the fix.
It smell like a fundamental issue for PC.

About kaffe in general - I built it initialy with DLL
but converted it to static libraries. I have this carzy
idea of running Kaffe on an embedded system which does
handle DLLs. In addition I will ROMize all the class
libraries to be loaded with the kaffe image as one
enchilada.

I'll probebly post progress along the way.

--Michael




---Daniel Veillard <veillard@rufus.w3.org> wrote:
>
> 
> >I was able to build kaffe under VC++ 5.0. It does
not
> >work yet but looks promissing. I have the following
> >problem:
> >
> >In initialization time while executing
initialization
> >of system class the vm gets into nested calls to
> >itself situation. In file kaffe.def:
> >
> >define_insn(INVOKESTATIC)
> >{
> >......
> >
> >#if defined(TRANSLATOR) && defined(HAVE_TRAMPOLINE))
> >  
>
>call_indirect_const(&METHOD_NATIVECODE(method_method());
> >#else
> >   call(tmp);
> >#endif
> >
> >........
> >}
> >
> >In my configuration the call(tmp) is what compiled
> >and, if you look at this macro it is, basicly,
calling
> >virtualMachine() which again gets back to
INVOKESTATIC
> >case and like that - forever until the program runs
> >out of stack.
> >
> >Any clue????
> 
>    Not really, did you try to compile the
interpreter or the JIT.
> BTW I'm very interested in a version of Kaffe in
the VC++ environment.
> Did you had to patch a lot of things ?
> 
> Daniel
> 
> 
> 
> 

_____________________________________________________________________
Sent by RocketMail. Get your free e-mail at http://www.rocketmail.com

From kaffe@kaffe.org  Wed Oct 29 18:05:42 1997
From: kaffe@kaffe.org (Aaron Coday)
Date: Wed, 29 Oct 1997 12:05:42 -0600
Subject: Kaffe bug on PC
References: <199710291609.LAA23851@rufus.w3.org>
Message-ID: <34577AF6.FF446DB9@cs.uiuc.edu>

Hi,

I have in the past worked on getting Kaffe running on win32/VC++, but
always run into a bunch of problems.  I think it would be very helpful
if when you get it working you could post some notes/diffs about
problems specific to win32/VC++.

thanks in advance.
-Aaron


Daniel Veillard wrote:
> 
> >I was able to build kaffe under VC++ 5.0. It does not
> >work yet but looks promissing. I have the following
> >problem:
> >
> >In initialization time while executing initialization
> >of system class the vm gets into nested calls to
> >itself situation. In file kaffe.def:
> >
> >define_insn(INVOKESTATIC)
> >{
> >......
> >
> >#if defined(TRANSLATOR) && defined(HAVE_TRAMPOLINE))
> >
> >call_indirect_const(&METHOD_NATIVECODE(method_method());
> >#else
> >   call(tmp);
> >#endif
> >
> >........
> >}
> >
> >In my configuration the call(tmp) is what compiled
> >and, if you look at this macro it is, basicly, calling
> >virtualMachine() which again gets back to INVOKESTATIC
> >case and like that - forever until the program runs
> >out of stack.
> >
> >Any clue????
> 
>    Not really, did you try to compile the interpreter or the JIT.
> BTW I'm very interested in a version of Kaffe in the VC++ environment.
> Did you had to patch a lot of things ?
> 
> Daniel

-- 
===============================================================
Aaron Coday				      a-coday@uiuc.edu
2234 DCL  MC 258  	http://www.students.uiuc.edu/~a-coday/
1304 W. Springfield
Urbana, IL  61801	"No matter where you go, there you are"
(W)217-244-7116
===============================================================

From kaffe@kaffe.org  Wed Oct 29 20:21:03 1997
From: kaffe@kaffe.org (Robert Bates)
Date: Wed, 29 Oct 1997 14:21:03 -0600
Subject: Possible licensing issue with JavaSoft?
Message-ID: <3.0.3.32.19971029142103.00947bf0@zeh.com>

I just got off the phone with QNX about porting Kaffe to QNX and using it
in a commercial product... They expressed concern over my using it on a
commercial platform, and even though Kaffe is a clean-room project, they
have heard rumblings from JavaSoft about quashing any VM running without a
JavaSoft license on a commercialized platform.

Has anyone got any info regarding the legalities behind these
fears/concerns?  Basically, I'm investigating using the Kaffe port on QNX
to build commercially-available servers, and would like to know if I'm
getting a bit optimistic (legally) about being able to sell the systems
with Kaffe loaded on them... QNX would love to know about it too, because
they are in the process of sinking serious $$$ into a JavaSoft license to
get a VM ported to run under QNX and Neutrino.

BTW, their's won't be available until mid '98, and they are actually urging
me to explore this avenue for an "immediate" Java solution... (I think they
have a vested interest in my product shipping ASAP... ;)

Thanks,
Rob
//////////////////////////////////////////////////////////////////////////
// Robert Bates
// Infosys Development Group
// Houston, Texas  USA
//
// rbates@iname.com
// http://home.att.net/~rpbates
//////////////////////////////////////////////////////////////////////////
#ifdef _LITIGATORS_PRESENT_
#include <std_disclaimer.h>
#endif

From kaffe@kaffe.org  Wed Oct 29 20:31:58 1997
From: kaffe@kaffe.org (John D. Gwinner)
Date: Wed, 29 Oct 1997 15:31:58 -0500
Subject: Kaffe bug on PC
Message-ID: <199710292019.UAA29778@home.virtual-pc.com>

Michael:
...
> it has to do with the difference between the compilers
> and the way they handle unions. The size of the slots
> union is 8 bytes. I see the pointer in the upper 4
> bytes in both the PC and Solaris. gcc takes the pointer
> correctly. VC++ takes the lower 4 bytes which are NULL.
> I'll address this problem today and broadcast the fix.
> It smell like a fundamental issue for PC.

Just to cover all bases, have you examined the packing (#pragma pack)?  I
don't think there is a guaranteed byte placement within a union, but I
could be wrong -- I'm just guessing.

Are you using INT64 for the 8 byte quantity to break out the 4 bytes? 
That might be one way to go.

		== John ==

From kaffe@kaffe.org  Wed Oct 29 20:50:02 1997
From: kaffe@kaffe.org (Rob Mayoff)
Date: Wed, 29 Oct 1997 14:50:02 -0600 (CST)
Subject: Possible licensing issue with JavaSoft?
In-Reply-To: <3.0.3.32.19971029142103.00947bf0@zeh.com>
Message-ID: <Pine.LNX.3.96.971029144658.20767D-100000@dqd.com>

+-- On Wed, 29 Oct 1997, Robert Bates wrote:
| I just got off the phone with QNX about porting Kaffe to QNX and using it
| in a commercial product... They expressed concern over my using it on a
| commercial platform, and even though Kaffe is a clean-room project, they
| have heard rumblings from JavaSoft about quashing any VM running without a
| JavaSoft license on a commercialized platform.

Read the copyright page of "The Java Language Specifications".  It
grants a license to implement the spec, but you have to do some things
like pass the test suites.

From kaffe@kaffe.org  Wed Oct 29 22:49:02 1997
From: kaffe@kaffe.org (Tony Kimball)
Date: Wed, 29 Oct 1997 16:49:02 -0600 (CST)
Subject: Possible licensing issue with JavaSoft?
References: <3.0.3.32.19971029142103.00947bf0@zeh.com>
Message-ID: <199710292249.QAA04668@compound.east.sun.com>

Quoth Robert Bates on Wed, 29 October:
: I just got off the phone with QNX about porting Kaffe to QNX and using it
: in a commercial product... They expressed concern over my using it on a
: commercial platform, and even though Kaffe is a clean-room project, they
: have heard rumblings from JavaSoft about quashing any VM running without a
: JavaSoft license on a commercialized platform.

Where have they heard this?

: Has anyone got any info regarding the legalities behind these
: fears/concerns? 

I'm not a lawyer but I've been sued in the past:-).  I am very much of
the opinion that if JavaSoft were to sue you for using Kaffe, you
would win a bunch of money in your countersuit.  Moreover, JavaSoft
people (not corporate officers) have told me that Java is an open
spec, and that such a suit is the farthest thing from their minds.
(Frankly, I think they'd lose a lot of crucial staff if they attacked
freeware.)  Finally, there are commerical products already in
existence using non-JavaSoft development kits and JVMs, from Japan,
and there is no sign of legal action from JavaSoft -- quite the
contrary, they seem to be actively cooperating on interface
specification issues for embedded systems.

On the otherhand, if you were to use the word "Java" you would be
inviting some sort of action, since they regard trademark dilution as
a serious threat.  "JV" or "J" are typical work-arounds, allowing you
to identify the functional characteristics of your product without
implying certifiable compliance.

: QNX would love to know about it too, because
: they are in the process of sinking serious $$$ into a JavaSoft license to
: get a VM ported to run under QNX and Neutrino.

That's the only way they can ever use the Java name, branding, or
logos, and have legal peace.  This has been JavaSoft's clearly
defended turf.

From kaffe@kaffe.org  Thu Oct 30 00:36:48 1997
From: kaffe@kaffe.org (Per Bothner)
Date: Wed, 29 Oct 1997 16:36:48 -0800
Subject: Possible licensing issue with JavaSoft?
In-Reply-To: Your message of "Wed, 29 Oct 1997 14:21:03 CST."
 <3.0.3.32.19971029142103.00947bf0@zeh.com>
Message-ID: <199710300035.QAA14937@cygnus.com>

> They expressed concern over my using it on a
> commercial platform, and even though Kaffe is a clean-room project, they
> have heard rumblings from JavaSoft about quashing any VM running without a
> JavaSoft license on a commercialized platform.

There are three separate issues:

a) Use of the trade-marked "Java" name.  In commercial use you would want
to be very careful when using "Java" without Sun's permission.  This is
an issue that came up when Sun proposed Java for ISO certification.
There is some Sun-approved work-around (something like "Java platform")
- look around java.sun.com.

2) Whether Sun would sue (or threaten to sue) over a commercial
clean-room re-implementation of Java, and if they did, whether they
would have a reasonable chance of prevailing.  I believe the answer
to these questions is no, and Cygnus is investing big bucks on that
belief.

3) Whether Kaffe is a clean-room implementation of Java.
Even though Kaffe is a clean-room re-implementation (or at least
close enough), the fact is that Kaffe is useless without Sun's
classes.zip.  And the license on classes.zip prevents its use
with Kaffe in a commercial context.  That is why a free replacement
for classes.zip is needed.  Cygnus will be re-implementing at
least the core classes.  (We have not looked at Kore or any
alternatives yet, so I cannot say to what extent we will
start with that.)  But we cannot re-implement all the features
in Sun's classes.zip in a timely manner (especially when Sun
keeps adding things like Java2D, Java3D, JDBC, etc etc), so
we hope "the Net" will help with the non-core classes.
(The core classes are too critical for us to wait for the net to
provide.)

	--Per Bothner
Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner

From kaffe@kaffe.org  Thu Oct 30 00:59:18 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Wed, 29 Oct 1997 16:59:18 -0800
Subject: Possible licensing issue with JavaSoft?
In-Reply-To: Your message of "Wed, 29 Oct 1997 16:36:48 PST."
 <199710300035.QAA14937@cygnus.com>
Message-ID: <199710300059.QAA20853@bach.marimba.com>


>a) Use of the trade-marked "Java" name.  In commercial use you would want
>to be very careful when using "Java" without Sun's permission.  This is
>an issue that came up when Sun proposed Java for ISO certification.
>There is some Sun-approved work-around (something like "Java platform")
>- look around java.sun.com.

What's the status of the "Unix" name?  Failing a Sun-approved 
work-around, I suppose that "J*va" would work, and be pretty useful, 
much like Un*x etc.

>2) Whether Sun would sue (or threaten to sue) over a commercial
>clean-room re-implementation of Java, and if they did, whether they
>would have a reasonable chance of prevailing.  I believe the answer
>to these questions is no, and Cygnus is investing big bucks on that
>belief.

I remember a tale of someone who put a Java logo on his homepage in 
the very early days of the language, to show his support.  But then 
Sun threatened to sue him...  I think that Sun's legal department has 
been quite stupidly aggressive in the past, and is probably still so.  
Whether or not they stand a chance of winning, they might try.  
Perhaps they were just bored back then, in which case we're all safe 
for now ;)

>3) Whether Kaffe is a clean-room implementation of Java.
>Even though Kaffe is a clean-room re-implementation (or at least
>close enough), the fact is that Kaffe is useless without Sun's
>classes.zip.

"The Net" plans to help out, I guess.  Whether they can keep up or not 
is another question, especially when Sun starts pulling a Microsoft 
and changing core classes gratuitously, as I expect they'll do when 
they feel comfortable about MS.  We will live in interesting times, 
indeed.

It would be kind of silly for any company _not_ to want a monopoly, 
wouldn't it?

-- 
benjamin@marimba.com            http://www.marimba.com/people/benjamin
PGP public key on my homepage, and finger -l bwpearre@cs.princeton.edu
1024/CC2761B9 97/2/11 E3 60 33 2F 1F 9D 17 60  DD 45 46 7E DF 1F 81 89 


From kaffe@kaffe.org  Thu Oct 30 02:06:52 1997
From: kaffe@kaffe.org (Michael Gesundheit)
Date: Wed, 29 Oct 1997 18:06:52 -0800 (PST)
Subject: Kaffe bug on PC
Message-ID: <19971030020652.24174.rocketmail@web2.rocketmail.com>






---"John D. Gwinner" <gwinner@northnet.org> wrote:
>
> Michael:
> ..
> > it has to do with the difference between the
compilers
> > and the way they handle unions. The size of the
slots
> > union is 8 bytes. I see the pointer in the upper 4
> > bytes in both the PC and Solaris. gcc takes the
pointer
> > correctly. VC++ takes the lower 4 bytes which are
NULL.
> > I'll address this problem today and broadcast the
fix.
> > It smell like a fundamental issue for PC.
> 
> Just to cover all bases, have you examined the
packing (#pragma pack)? 

No, under 0.9.2 the thread package does not require
assembly. setjmp/longjmp is used.
 I
> don't think there is a guaranteed byte placement
within a union, but I
> could be wrong -- I'm just guessing.
> 
> Are you using INT64 for the 8 byte quantity to
break out the 4 bytes? 

Yes.

> That might be one way to go.

The problem as I see it at this point is not a union
issue. I realized today that the problem comes from
the initial GETSTATIC opcode processing. The
move_ref_const macro moves the pointer
field->info.addr to tmp which is a slots pointer then
the macro load_ref move the contant, which is 0 (!!) to
the stack. Later on the bytecode INVOKEVIRTUAL takes
this stack entry as a pointer and crashes. At this 
point I think the bug is in initializing my 
HelloWorldApp class. This missing pointer should be 
set when the class is loaded and processed. I hope
that the Solaris I have next to the PC will help
tracing the class initialization.

--Michael
macro moves a field->info.addr pointer which points
to 0. The following move_ref_const

_____________________________________________________________________
Sent by RocketMail. Get your free e-mail at http://www.rocketmail.com

From kaffe@kaffe.org  Thu Oct 30 12:50:22 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Thu, 30 Oct 1997 13:50:22 +0100
Subject: Alpha linux hangs when using javac
Message-ID: <199710301250.NAA01488@marvinii.link.no>


>Submitter-Id:  kaffe-user
>Originator:    Petter Reinholdtsen
>Organization:  Telenor Link
>Confidential:  no
>Synopsis:      kaffe hangs when running javac
>Severity:      critical
>Priority:      high
>Category:      kaffe
>Class:         sw-bug
>Release:	0.81
>Environment:
        System: Linux xenu 2.0.30 #6 Tue May 13 19:31:06 CEST 1997 alpha unknown
        Architecture: alpha
        Machine: DEC
>Description:
	kaffe hangs when trying to compile something using javac.  It
	uses 90% CPU, but is never done.
>How-To-Repeat:
	After getting kaffe to compile on Linux Alpha (had to add
	-Wl,-rpath,/usr/local/lib to the linking of kaffe), I tried
	running javac to compile test/general/RandomTest.java.  I had
	to add CLASSPATH=/usr/local/share/kaffe/classes.zip and
	LD_LIBRARY_PATH=/usr/local/lib to environment to get it to run
	at all.  The compiler never finishes.
>Fix:
	Don't know.

From kaffe@kaffe.org  Thu Oct 30 14:01:19 1997
From: kaffe@kaffe.org (David Scott)
Date: Thu, 30 Oct 1997 09:01:19 -0500 (EST)
Subject: Possible licensing issue with JavaSoft?
In-Reply-To: <199710300059.QAA20853@bach.marimba.com>
Message-ID: <Pine.LNX.3.95.971030084932.11373D-100000@studmuffin.ganymede.com>


On Wed, 29 Oct 1997 benjamin@marimba.com wrote:
 
> I remember a tale of someone who put a Java logo on his homepage in 
> the very early days of the language, to show his support.  But then 
> Sun threatened to sue him...  I think that Sun's legal department has 
> been quite stupidly aggressive in the past, and is probably still so.  
> Whether or not they stand a chance of winning, they might try.  
> Perhaps they were just bored back then, in which case we're all safe 
> for now ;)

My take on this issue is this: Sun is being very careful in defining its
legal position with regard to Java(tm) because of Microsoft, whom Sun
doesn't want mucking around with Java the way they muck around with
everything else. The thing is, for them to defend their turf against the
Evil Empire, they have to enforce some silly-looking boilerplate even
among more friendly folk. 

For example, it appears that O'Reilly Associates has recently heard from
Sun's lawyers about using the Java name without the (tm) on the covers of
their excellent Java books. This is particularly bizarre when you consider
how Java-friendly ORA really is. 

For the time being, I think Sun should be given very wide berth on this
issue, until things sort themselves out a bit.  I find their lawsuit
against Microsoft very interesting.  One way to support them is simply to
play along. 

David Scott
Montreal

From kaffe@kaffe.org  Thu Oct 30 14:21:35 1997
From: kaffe@kaffe.org (Otto Hammersmith)
Date: Thu, 30 Oct 1997 09:21:35 -0500
Subject: Alpha linux hangs when using javac
In-Reply-To: <199710301250.NAA01488@marvinii.link.no>; from pere@td.org.uit.no on Thu, Oct 30, 1997 at 01:50:22PM +0100
References: <199710301250.NAA01488@marvinii.link.no>
Message-ID: <19971030092135.54643@redhat.com>

FWIW, something similar happens on an intel with glibc 2
running... I'd bet that it's glibc and not a 32-64 bit problem.  I'm
intending to take a look at it in the next couple of days, I'll
certainly forward a patch if I come up with anything.

On Thu, Oct 30, 1997 at 01:50:22PM +0100, pere@td.org.uit.no wrote:
> 
> >Submitter-Id:  kaffe-user
> >Originator:    Petter Reinholdtsen
> >Organization:  Telenor Link
> >Confidential:  no
> >Synopsis:      kaffe hangs when running javac
> >Severity:      critical
> >Priority:      high
> >Category:      kaffe
> >Class:         sw-bug
> >Release:	0.81
> >Environment:
>         System: Linux xenu 2.0.30 #6 Tue May 13 19:31:06 CEST 1997 alpha unknown
>         Architecture: alpha
>         Machine: DEC
> >Description:
> 	kaffe hangs when trying to compile something using javac.  It
> 	uses 90% CPU, but is never done.
> >How-To-Repeat:
> 	After getting kaffe to compile on Linux Alpha (had to add
> 	-Wl,-rpath,/usr/local/lib to the linking of kaffe), I tried
> 	running javac to compile test/general/RandomTest.java.  I had
> 	to add CLASSPATH=/usr/local/share/kaffe/classes.zip and
> 	LD_LIBRARY_PATH=/usr/local/lib to environment to get it to run
> 	at all.  The compiler never finishes.
> >Fix:
> 	Don't know.
> 

-- 
					-Otto.

From kaffe@kaffe.org  Thu Oct 30 20:20:50 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Thu, 30 Oct 1997 20:20:50 GMT
Subject: Finally...help with
Message-ID: <199710302020.UAA13895@home.virtual-pc.com>

building a great Web Site at an unbelievably low price.
from: sitehelp@worldwidepromo.com

Many people would like to put their business on the Web but are
frustrated trying to do it themselves or getting someone else to do it
all at a reasonable price. 

If you want to do it yourself and just need a road map for how it
is done. We can help.

If you have been working on a site and it isn't coming together the way
you would have liked. We can help.

If you don't have a clue about Web Site design and don't want anything
to do with it, but want a great site. We can help.

We can show you examples of sites we have done and explain how we did
them or we can take your idea and put it on the Web without anything
from you but input.

We offer custom sites with high quality graphics from our extensive
graphics libraries. Together with copy writing and design capabilities
we combine the use of the latest advances in the media to produce sites
that are interesting, informative and captivating. 

Please reply at:        sitehelp@worldwidepromo.com
..so we can help you get the job done.







From kaffe@kaffe.org  Thu Oct 30 13:32:25 1997
From: kaffe@kaffe.org (spinny)
Date: Thu, 30 Oct 1997 13:32:25 +0000 (GMT)
Subject: Possible licensing issue with JavaSoft?
In-Reply-To: <Pine.LNX.3.95.971030084932.11373D-100000@studmuffin.ganymede.com>
Message-ID: <Pine.LNX.3.96.971030132743.3922A-100000@borovnia.spinny.com>



On Thu, 30 Oct 1997, David Scott wrote:

> 
> On Wed, 29 Oct 1997 benjamin@marimba.com wrote:
>  
> > I remember a tale of someone who put a Java logo on his homepage in 
> > the very early days of the language, to show his support.  But then 
> > Sun threatened to sue him...  I think that Sun's legal department has 
> > been quite stupidly aggressive in the past, and is probably still so.  
> > Whether or not they stand a chance of winning, they might try.  
> > Perhaps they were just bored back then, in which case we're all safe 
> > for now ;)
> 
> My take on this issue is this: Sun is being very careful in defining its
> legal position with regard to Java(tm) because of Microsoft, whom Sun
> doesn't want mucking around with Java the way they muck around with
> everything else. The thing is, for them to defend their turf against the
> Evil Empire, they have to enforce some silly-looking boilerplate even
> among more friendly folk. 
> 
> For example, it appears that O'Reilly Associates has recently heard from
> Sun's lawyers about using the Java name without the (tm) on the covers of
> their excellent Java books. This is particularly bizarre when you consider
> how Java-friendly ORA really is. 
> 
> For the time being, I think Sun should be given very wide berth on this
> issue, until things sort themselves out a bit.  I find their lawsuit
> against Microsoft very interesting.  One way to support them is simply to
> play along. 
> 
> David Scott
> Montreal
> 

I couldn't agree more.  It would be awful if they sat around and judged
the individual intentions of all the various organizations that are in
some way involved in Java development and then, based on these
calculations, decided which organizations to prosecute.  A unified
structure works much better.  And this is a crucial juncture in the life
of Java. I do wonder how things might be different without the spectre of
Microsoft lurking about. 

garner

From kaffe@kaffe.org  Thu Oct 30 23:18:44 1997
From: kaffe@kaffe.org (Gregor Hoffleit)
Date: Fri, 31 Oct 97 00:18:44 +0100
Subject: kaffe for NEXTSTEP, was: Re: NetBSD-m68k-0.9.1 compile problem
In-Reply-To: <yam7238.1178.124108720@faui01.informatik.uni-erlangen.de>
References: <yam7238.1178.124108720@faui01.informatik.uni-erlangen.de>
Message-ID: <9710302318.AA21365@thefly.mathi.uni-heidelberg.de>

--NeXT-Mail-1425441485-1
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: quoted-printable

Matthias Hopf wrote:
    > Don't know why I stand in NOTES/HELPERS as NeXT porter [Hier war =
urspr=FCnglich eine Anlage beigef=FCgt]
   =20
    This is still correct, though. I thought I should send a message to Tim
    after my final exams (tuesday), but then, I can do it immedeately...
   =20
    Tim, could you please remove me as NeXT porter in NOTES/HELPERS? I'm
    repeatedly getting questions about the NeXT port, and I can't help those
    folks. I don't know, who's done the NeXT port, either...

I browsed my archives and found the following:

- The initial port to NEXTSTEP on m68k was done by Ralf E. Stranzenbach =
<ralf@reswi.ruhr.de> based on Matthias' Amiga code
- The port to NEXTSTEP on i386 and sparc was done by me (Gregor Hoffleit =
<flight@mathi.uni-heidelberg.de>) based on resp. code for FreeBSD/Solaris =
and with help and bugfixes from numerous other people.

Since I seem to be the only one actively working on kaffe for NEXTSTEP =
nowadays (speak up please!), Tim can feel free to add my name as "Project =
leader" for Kaffe on NEXTSTEP, i.e. for NOTES/HELPERS:

Gregor Hoffleit <flight@mathi.uni-heidelberg.de>    NEXTSTEP package of =
Kaffe


Furthermore, KAFFE 0.9.x DOESN'T WORK ON ANY NEXTSTEP platform, it's broken. =
You might want to add that to NOTES/BUGS or whereever for the next release.

	Gregor


---
|  Gregor Hoffleit     admin MATHInet / contact HeidelNeXT |
| MAIL: Mathematisches Institut   PHONE: (49)6221 56-5771 |
|       INF 288, 69120 Heidelberg / Germany  FAX: 56-3812 |
| EMAIL: flight@mathi.uni-heidelberg.de (NeXTmail)        |

--NeXT-Mail-1425441485-1
Content-Type: text/enriched; charset=iso-8859-1
Content-Transfer-Encoding: quoted-printable

Matthias Hopf wrote:

    > Don't know why I stand in NOTES/HELPERS as NeXT porter <

   =20

    This is still correct, though. I thought I should send a message to Tim

    after my final exams (tuesday), but then, I can do it immedeately...

   =20

    Tim, could you please remove me as NeXT porter in NOTES/HELPERS? I'm

    repeatedly getting questions about the NeXT port, and I can't help those

    folks. I don't know, who's done the NeXT port, either...


I browsed my archives and found the following:


- The initial port to NEXTSTEP on m68k was done by Ralf E. Stranzenbach =
<<ralf@reswi.ruhr.de> based on Matthias' Amiga code

- The port to NEXTSTEP on i386 and sparc was done by me (Gregor Hoffleit =
<<flight@mathi.uni-heidelberg.de>) based on resp. code for FreeBSD/Solaris =
and with help and bugfixes from numerous other people.


Since I seem to be the only one actively working on kaffe for NEXTSTEP =
nowadays (speak up please!), Tim can feel free to add my name as "Project =
leader" for Kaffe on NEXTSTEP, i.e. for NOTES/HELPERS:


Gregor Hoffleit <<flight@mathi.uni-heidelberg.de>    NEXTSTEP package of =
Kaffe



Furthermore, KAFFE 0.9.x DOESN'T WORK ON ANY NEXTSTEP platform, it's broken. =
You might want to add that to NOTES/BUGS or whereever for the next release.


	Gregor



---

|
<bigger>=20
</bigger><smaller> Gregor Hoffleit     admin MATHInet / contact HeidelNeXT |

| MAIL: Mathematisches Institut   PHONE: (49)6221 56-5771 |

|       INF 288, 69120 Heidelberg / Germany  FAX: 56-3812 |

| EMAIL: flight@mathi.uni-heidelberg.de (NeXTmail)        |

</smaller>

--NeXT-Mail-1425441485-1--

From kaffe@kaffe.org  Fri Oct 31 00:01:45 1997
From: kaffe@kaffe.org (Aleksey Sudakov)
Date: Thu, 30 Oct 97 19:01:45 -0500
Subject: kaffe for NEXTSTEP, was: Re: NetBSD-m68k-0.9.1 compile problem
In-Reply-To: <9710302318.AA21365@thefly.mathi.uni-heidelberg.de>
References: <yam7238.1178.124108720@faui01.informatik.uni-erlangen.de>
 <9710302318.AA21365@thefly.mathi.uni-heidelberg.de>
Message-ID: <9710310001.AA01346@rkinc.com>

<nofill>: Furthermore, KAFFE 0.9.x DOESN'T WORK ON ANY NEXTSTEP platform, it's
: broken. You might want to add that to NOTES/BUGS or whereever for the next
: release.
</nofill>

And perhaps better explain what are these trampoline functions for and why
they are not recognized by linker? Any mach assembler hacker around who
could help resolving the issue?


Aleksey

From kaffe@kaffe.org  Fri Oct 31 03:34:27 1997
From: kaffe@kaffe.org (John D. Gwinner)
Date: Thu, 30 Oct 1997 22:34:27 -0500
Subject: Kaffe bug on PC
Message-ID: <199710311521.PAA03227@home.virtual-pc.com>

Michael:

> ---"John D. Gwinner" <gwinner@northnet.org> wrote:

> > Just to cover all bases, have you examined the
> packing (#pragma pack)? 
> 
 No, under 0.9.2 the thread package does not require
> assembly. setjmp/longjmp is used.

? #pragma pack is useful even if you aren't using assembly;
I'm not sure what that has to do with setjmp/longjmp.  Maybe I'm not
following.  If you are reading 0's in structures that should have value,
this is often because the #pragma pack is wrong.  I've had problems with
this in the past.  (setting pack is a bad thing of course, but at least an
Intel machine doesn't GPF when a structure isn't aligned, but it does slow
down.).


> The problem as I see it at this point is not a union
> issue. I realized today that the problem comes from
> the initial GETSTATIC opcode processing. The
> move_ref_const macro moves the pointer
> field->info.addr to tmp which is a slots pointer then
> the macro load_ref move the contant, which is 0 (!!) to
> the stack. Later on the bytecode INVOKEVIRTUAL takes
> this stack entry as a pointer and crashes. At this 
> point I think the bug is in initializing my 
> HelloWorldApp class. This missing pointer should be 
> set when the class is loaded and processed. I hope
> that the Solaris I have next to the PC will help
> tracing the class initialization.

Hmm ... you mean the pointer is written to tmp properly, then it suddenly
becomes 0? Sounds like a wild pointer somewhere.  Set a break point on
that memory address and see what changes it to 0.  OR -- trace the macro
to make sure it's doing what it's supposed to.

I'm probably out of juice on this one, thought I could help.

		== John ==

From kaffe@kaffe.org  Fri Oct 31 18:10:18 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Fri, 31 Oct 1997 13:10:18 -0500
Subject: Important Kaffe mailing-list informations !
Message-ID: <199710311810.NAA14673@rufus.w3.org>

  Hi all,

  ====================================================
  This message has important informations about Kaffe
  mailing-list, please consider saving it for later use
  ====================================================

Remember, the mailing-list is set-up so that replying to
a message issued from the list goes back to the list.
PLEASE, take care of this !!!

  ====================================================
      the mailing list has moved, it's now served by 
               majordomo@rufus.w3.org
  ====================================================

This machine is in MIT domain, Boston, USA.
So to subscribe :

echo "subscribe" | mail majordomo@rufus.w3.org

to unsubscribe :

echo "unsubscribe" | mail majordomo@rufus.w3.org

However to send to the list use the kaffe@kaffe.org 
address, it didn't change.

  ====================================================
  Don't send unsubscribe messages to the list, use the
         majordomo@rufus.w3.org for admin message !
  ==================================================== 

 please allow a minimum amount of time for your request to
be processed, especially unsubscribing, because people always
unsubscribe from accounts different from where they are registered
(Murphy's Law applied to mailing list administration).
So I need to do it by hand !

  Considering spams and error messages forwarded to the mailing
list there is nothing (or very few things) I can do to avoid
them, please bear with the state of the art :-\ ... I just can't
fix remote broken SMTP agents. 
  If you have proven known way of improving things I would be
happy to heard from you about it.

  ======================================================
      The mailing list has an archives, read them !
        http://rufus.w3.org/tools/Kaffe/messages
      search from : http://rufus.w3.org/tools/Kaffe/
  ======================================================

  I will try to do my best to keep the quality of service to
it's maximum (I guess the delivery speed already improved by a
factor of 10), but this also need help from the people using
the list. Remember that a junk mail sent to it irritate around
600 persons !

  Thank you for having read this message, please keep it,

    yours,

  Daniel

-- 
Daniel.Veillard@w3.org | MIT/LCS   Room NE43-344 | Today's Bookmarks :
Tel : +1.617.253.5884  | 545 Technology Square   | Linux, WWW, Java,
Fax : +1.617.258.5999  | Cambridge, MA 02139 USA | badminton, Kaffe,
http://www.w3.org/People/W3Cpeople.html#Veillard | HTTP-NG and Amaya.


From kaffe@kaffe.org  Fri Oct 31 19:28:06 1997
From: kaffe@kaffe.org (Michael R Cook)
Date: 31 Oct 1997 14:28:06 -0500
Subject: Important Kaffe mailing-list informations !
In-Reply-To: Daniel.Veillard@w3.org's message of "Fri, 31 Oct 1997 13:10:18 -0500"
References: <199710311810.NAA14673@rufus.w3.org>
Message-ID: <r4yb39zq9l.fsf@cognex.com>

>>>>> "DV" == Daniel Veillard <Daniel.Veillard@w3.org> writes:

 DV> Remember, the mailing-list is set-up so that replying to
 DV> a message issued from the list goes back to the list.

Incidentally, that kind of behavior is generally considered to be
wrong: the list remailer should not be adding the Reply-to header.

In fact, the Gnus mail/news reader has special support for such
mailing lists.  The manual for Gnus says:

|`broken-reply-to'
|     Elements like `(broken-reply-to . t)' signal that `Reply-To'
|     headers in this group are to be ignored.  This can be useful if
|     you're reading a mailing list group where the listserv has inserted
|     `Reply-To' headers that point back to the listserv itself.  This is
|     broken behavior.  So there!

Michael.

From kaffe@kaffe.org  Fri Oct 31 19:46:59 1997
From: kaffe@kaffe.org (Per Bothner)
Date: Fri, 31 Oct 1997 11:46:59 -0800
Subject: Possible licensing issue with JavaSoft?
In-Reply-To: Your message of "Wed, 29 Oct 1997 16:59:18 PST."
 <199710300059.QAA20853@bach.marimba.com>
Message-ID: <199710311945.LAA06297@cygnus.com>

> It would be kind of silly for any company _not_ to want a monopoly, 
> wouldn't it?

Nope.  It's kind of silly to want a monopoly.  It invites all kinds
of unwelcome attention (legally, upstart comptetitors, potshots from
journalists).  It also tends to make you too complacent, which is
dangerous in the long run.

What you want to be is the largest supplier, the one with the
brand-name recogniztion, and that nobody gets fired for buying.
E.g. Intel and Microsoft.

	--Per Bothner
Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner


From kaffe@kaffe.org  Fri Oct 31 21:25:32 1997
From: kaffe@kaffe.org (Clyde Smith-Stubbs)
Date: Sat, 1 Nov 1997 07:25:32 +1000
Subject: Possible licensing issue with JavaSoft?
In-Reply-To: <199710311945.LAA06297@cygnus.com>; from Per Bothner on Fri, Oct 31, 1997 at 11:46:59AM -0800
References: <199710300059.QAA20853@bach.marimba.com> <199710311945.LAA06297@cygnus.com>
Message-ID: <19971101072532.17804@htsoft.com>

On Fri, Oct 31, 1997 at 11:46:59AM -0800, Per Bothner wrote:
> Nope.  It's kind of silly to want a monopoly.  It invites all kinds
> of unwelcome attention (legally, upstart comptetitors, potshots from

Having multiple vendors also grows the total market size - often
resulting in more for each vendor than they could have expected if
they had to do that all themselves.

-- 
Clyde Smith-Stubbs               |            HI-TECH Software
Email: clyde@htsoft.com          |          Phone            Fax
WWW:   http://www.htsoft.com/    | USA: (408) 490 2885  (408) 490 2885
PGP:   finger clyde@htsoft.com   | AUS: +61 7 3354 2411 +61 7 3354 2422
---------------------------------------------------------------------------
ANSI C for the PIC! Now shipping! See www.htsoft.com for more info.

From kaffe@kaffe.org  Fri Oct 31 21:41:44 1997
From: kaffe@kaffe.org (Lee Elson)
Date: Fri, 31 Oct 1997 13:41:44 -0800
Subject: Linux and biss-awt
Message-ID: <9710311341.ZM20919@magus.jpl.nasa.gov>

I'm a new user and am having a problem setting things up. The error
message I get is:
java.lang.UnsatisfiedLinkError: no bissawt in shared library path
        at java/lang/Throwable.<init>(line unknown, pc 0x836a90f)
        at java/lang/Error.<init>(line unknown, pc 0x836aa61)
        at java/lang/LinkageError.<init>(line unknown, pc 0x836a9e9)
        at java/lang/UnsatisfiedLinkError.<init>(line unknown, pc
0x836a971)
        at java/lang/Runtime.loadLibrary(line unknown, pc 0x8368b27)
        at java/lang/System.loadLibrary(line unknown, pc 0x836a721)
        at biss/awt/kernel/NativeLib.<clinit>(line unknown, pc 0x8366e16)
.
.
.
.
Here's what I have in /usr/local/lib:
libkaffe_bissawt.so.0.87  libnative.so.0.92         libzip.so.0.92
libagent.so@              libkaffevm.so@            libnet.so@
libagent.so.0.92          libkaffevm.so.0.92        libnet.so.0.92
libkaffe_bissawt.so@      libnative.so@             libzip.so@

I have LD_LIBRARY_PATH set to /usr/local/lib and I've run ldconfig. Is
libkaffe_bissawt.so.0.87 the latest version? Anyone see any other problems
which cause the above error? Any help is appreciated.

Lee Elson (elson@magus.jpl.nasa.gov)

From kaffe@kaffe.org  Wed Oct  1 14:42:59 1997
From: kaffe@kaffe.org (Mikael Ståldal)
Date: Wed, 01 Oct 1997 14:42:59 +0100
Subject: Porting Kaffe to OS/2
Message-ID: <jNlM0EpfhKYU092yn@ingen.reklam.nada.kth.se>

I'm trying to port Kaffe to OS/2 (an i386 platform) using the EMX compiler
(GCC port). I have encountered some problems.

Some different header files differ only in case, and since the file system
in OS/2 isn't case sensitive this causes clashes. String.h clashes with the
standard string.h and System.h clashes with Kaffe's system.h.

Make doesn't find machine.c in the interp subdirectory, how is it supposed
to find it there? I'm using GNU Make 3.75. When I copied it to kaffevm it
works. The same problem with the package directoris in packages.

When I copied files from the packages, things broke due to "../../.." style
#include directives that I hade to change manually. Try to avoid using ".."
in #include directives.

I get the following error:

make[1]: Entering directory `/UNIX/kaffe/kaffe/kaffevm'
gcc -g -O -Zexe -DNO_SHARED_LIBRARIES -I. -I. -I./../../config -I../../config -I../../include -I./../../include -DINTERPRETER -I./intrp   -DKVER=\"0.91\" -c  machine.c
machine.c: In function `virtualMachine':
machine.c:141: structure has no member named `exception_table_len'
machine.c:152: structure has no member named `ins'
machine.c:178: structure has no member named `code'
machine.c:182: structure has no member named `codelen'
make[1]: *** [machine.o] Error 1
make[1]: Leaving directory `/UNIX/kaffe/kaffe/kaffevm'
make: *** [all] Error 1

When I look at the structure, it has no exception_table_len member. How is
this supposed to works?

-- 
/****************************************************************\
* You have just read a message from Mikael Ståldal.              *
*                                                                *
* Remove "ingen.reklam." from the address before mail replying.  *
\****************************************************************/

From kaffe@kaffe.org  Thu Oct  2 01:06:44 1997
From: kaffe@kaffe.org (Ram Madduluri)
Date: Wed, 01 Oct 1997 18:06:44 -0600
Subject: Kaffe runtime question
Message-ID: <s4328dbf.005@novell.com>

Hello,

I have ported Kaffe 0.9.1 to OSF/1 on x86 platform. The initial
HelloWorldApp test passed. However, when I run a sample server application -
UDPEchoServer, I get the following error:
	$ kaffe UDPEchoServer
	My IP: 127.0.0.1
	My port: 6666
	Failed to locate native function:
        	java/net/PlainDatagramSocketImpl.datagramSocketCreate()V
	java.lang.UnsatisfiedLinkError
        	at java/net/DatagramSocket.create(133)
        	at java/net/DatagramSocket.<init>(114)
        	at java/net/DatagramSocket.<init>(100)
        	at UDPServer.main(41)

The PlainDatagramSocketImpl class is in the
/usr/local/share/kaffe/classes.zip file. Am I missing something here? I can
run some of the other tests just fine (DemoServer, DemoClient, ..). Is there
some configuration stuff that I messed up??

Any pointers would be greatly appreciated.

Regards,

Ram
--
Ram Madduluri                                                  E-Mail:
ram@Novell.COM
Corporate Research                                          Voice:  
801-222-5361
Novell Inc.                                                          Fax:   
  801-222-5177
1555 North Technology Way
Orem, UT 84097

From kaffe@kaffe.org  Thu Oct  2 05:33:54 1997
From: kaffe@kaffe.org (Orville R. Weyrich)
Date: Wed, 1 Oct 1997 21:33:54 -0700 (MST)
Subject: Porting Kaffe to OS/2
In-Reply-To: <jNlM0EpfhKYU092yn@ingen.reklam.nada.kth.se>
Message-ID: <Pine.LNX.3.91.971001210131.6624A-100000@dopey.weyrich.com>

On Wed, 1 Oct 1997, Mikael Ståldal wrote:

> I'm trying to port Kaffe to OS/2 (an i386 platform) using the EMX compiler
> (GCC port). I have encountered some problems.
> 
> Some different header files differ only in case, and since the file system
> in OS/2 isn't case sensitive this causes clashes. String.h clashes with the
> standard string.h and System.h clashes with Kaffe's system.h.

My suggestion is to place all kaffe include files in a directory named
kaffe, such that the kaffe directory is in your INCLUDE path.  This means
that you will have to modify the source code to:

	#include <string.h>

and

	#include <kaffe/String.h>

It is less work if you only do this for the files which conflict, but it
is cleaner if you move all the kaffe headers to this location and tweak the
source code appropriately.  In my opinion, this would add value to the 
source code base for all users, since it would make clear which includes
are for system headers, and which are for kaffe headers.

I just downloaded kaffe myself, and haven't looked at the code yet,
so some other variation may be more consistent with the existing
organization, but this should get you started.  I use OS/2 and would
love to have kaffe on OS/2, so give me a shout if I can be of any
more help :-)

> 
> Make doesn't find machine.c in the interp subdirectory, how is it supposed
> to find it there? I'm using GNU Make 3.75. When I copied it to kaffevm it
> works. The same problem with the package directoris in packages.

Is there a configuration script that generates machine.c based on the
environment?
 
> When I copied files from the packages, things broke due to "../../.." style
> #include directives that I hade to change manually. Try to avoid using ".."
> in #include directives.

I am afraid that there is no general solution for this portability problem,
absolute paths also have their problems.  Perhaps a perl script containing
something like

	if ( $line =~ /^#include/ )
	{
		$line =~ s/\.\.\//..\\/;
	}

might be the easiest to do over all your source files.
(i.e. find all include lines and change ../ to ..\).

> 
> I get the following error:
> 
> make[1]: Entering directory `/UNIX/kaffe/kaffe/kaffevm'
> gcc -g -O -Zexe -DNO_SHARED_LIBRARIES -I. -I. -I./../../config 
-I../../config -I../../include -I./../../include -DINTERPRETER -I./intrp   -DKVER=\"0.91\" -c  machine.c
> machine.c: In function `virtualMachine':
> machine.c:141: structure has no member named `exception_table_len'
> machine.c:152: structure has no member named `ins'
> machine.c:178: structure has no member named `code'
> machine.c:182: structure has no member named `codelen'
> make[1]: *** [machine.o] Error 1
> make[1]: Leaving directory `/UNIX/kaffe/kaffe/kaffevm'
> make: *** [all] Error 1
> 
> When I look at the structure, it has no exception_table_len member. How is
> this supposed to works?
> 

I don't know.  It may be due to using the wrong header file.

===================================================================
Orville R. Weyrich, Jr.                 Weyrich Computer Consulting
mailto:orville@weyrich.com                   http://www.weyrich.com
-------------------------------------------------------------------

From kaffe@kaffe.org  Thu Oct  2 13:14:41 1997
From: kaffe@kaffe.org (David Turner)
Date: Thu, 2 Oct 1997 13:14:41 +0100 (MET DST)
Subject: Porting Kaffe to OS/2
In-Reply-To: <jNlM0EpfhKYU092yn@ingen.reklam.nada.kth.se> from "Mikael Ståldal" at Oct 1, 97 02:42:59 pm
Message-ID: <199710021114.NAA08384@messiaen.enst.fr>

Hi Mikael

Mikael Ståldal said:
> 
> machine.c: In function `virtualMachine':
> machine.c:141: structure has no member named `exception_table_len'
> machine.c:152: structure has no member named `ins'
> machine.c:178: structure has no member named `code'
> machine.c:182: structure has no member named `codelen'
> make[1]: *** [machine.o] Error 1
> make[1]: Leaving directory `/UNIX/kaffe/kaffe/kaffevm'
> make: *** [all] Error 1
> 
> When I look at the structure, it has no exception_table_len member. How is
> this supposed to works?
>

It's a famous bug in the 0.9.1 release (it's not EMX specific ;-)
The fix has been released in the mail archive, but I'll give it here
again, from memory :

  1. replace 'method->codelen' by 'method->c.bcode.codelen'  (line 182)
     and     'method->code'    by 'method->c.bcode.code'     (line 178)

     ( I don't remember if it's "method" or "meth" or "m", but it's
       a pointer to a method structure )

  2. replace 'exception_table_len > 0' by 'exception_table != 0' (line 141)

  3. define a local variable "char* str" and replace
     line 152 with :

      "str = meth->signature->data;
       idx = sizeofSig( &str, false ) + ( .... & ..... ? ... : ... )'
                                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                                          same as before..

  Then recompile, it should work OK now.. ( well, there are still some
  bugs in there ;-)

Good luck, you'll need it..

- David

From kaffe@kaffe.org  Thu Oct  2 17:59:03 1997
From: kaffe@kaffe.org (Tony Kimball)
Date: Thu, 2 Oct 1997 11:59:03 -0500 (CDT)
Subject: Porting Kaffe to OS/2
References: <jNlM0EpfhKYU092yn@ingen.reklam.nada.kth.se>
 <Pine.LNX.3.91.971001210131.6624A-100000@dopey.weyrich.com>
Message-ID: <199710021659.LAA06487@compound.east.sun.com>

Quoth Orville R. Weyrich on Wed, 1 October:
:  
: > When I copied files from the packages, things broke due to "../../.." style
: > #include directives that I hade to change manually. Try to avoid using ".."
: > in #include directives.
: 
: I am afraid that there is no general solution for this portability problem,
: absolute paths also have their problems. 

-I fixes it everywhere.

From kaffe@kaffe.org  Thu Oct  2 18:49:10 1997
From: kaffe@kaffe.org (James A. Hillyerd)
Date: Thu, 2 Oct 1997 10:49:10 -0700 (PDT)
Subject: Next Release?
Message-ID: <Pine.LNX.3.95.971002104801.6765B-100000@animal.blarg.net>

Any guesses on when the next release of Kaffe is coming out?  This list
has been pretty quiet lately. =)

-james

[  James A. Hillyerd (JH2162) - james@blarg.net - Web Developer  ]
[  http://www.blarg.net/~james/   http://www.hyperglyphics.com/  ]
[ 1024/B11C3751 CA 1C B3 A9 07 2F 57 C9  91 F4 73 F2 19 A4 C5 88 ]

From kaffe@kaffe.org  Sat Oct  4 04:49:11 1997
From: kaffe@kaffe.org (Michael Hirsch)
Date: Fri, 3 Oct 1997 23:49:11 -0400 (EDT)
Subject: java.lang.ClassNotFoundException:HelloWorldApp
Message-ID: <199710040349.XAA22083@cssun.mathcs.emory.edu>

I have the same problem that Bryan Felix reported.

I tried to follow the README instructions exactly.  I got all four
tarballs: 
kaffe-0.9.1.tgz 
kaffe-0.9.1-package-javasoft.com.tgz
kaffe-0.9.1-package-biss-net.com.tgz and
kaffe-0.9.1-package-epfl.ch.tgz 
and unpacked them in the same directory.

./configure worked fine, as did make and make install.

But now:
[34] test> echo $CLASSPATH $KAFFEHOME
.:/usr/local/share/kaffe/classes.zip:/usr/local/share/kaffe/biss.zip /usr/local/share/kaffe
[35] test> pwd
/usr/local/Uploads/kaffe-0.9.1/test
[36] test> kaffe HelloWorldApp
java.lang.ClassNotFoundException: HelloWorldApp
[37] test> 

I tried to do exactly as it says.  . is in the CLASSPATH.
I did not compile HelloWorldApp.fava into a .class file because 1) I
don't know how, exactly, and 2) the README doesn't say to.

However, if I try
[37] test> javac HelloWorldApp.java
my screen fills with this error message:
java.lang.NullPointerException
	at java/io/File.list(line unknown, pc 0x8520a6c)
	at sun/tools/java/ClassPathEntry.getFiles(line unknown, pc 0x8532197)
	at sun/tools/java/ClassPath.getFile(line unknown, pc 0x852ddb9)
	at sun/tools/java/ClassPath.getFile(line unknown, pc 0x8530bd6)
	at sun/tools/java/Package.getBinaryFile(line unknown, pc 0x853477a)
	at sun/tools/java/Package.classExists(line unknown, pc 0x852083b)
	at sun/tools/javac/BatchEnvironment.classExists(line unknown, pc 0x852c650)
	at sun/tools/java/Environment.classExists(line unknown, pc 0x8530369)
	at sun/tools/java/Environment.classExists(line unknown, pc 0x8530369)
	at sun/tools/java/Imports.importable(line unknown, pc 0x851ee55)
	at sun/tools/java/Imports.resolve(line unknown, pc 0x852d2db)
	at sun/tools/java/ImportEnvironment.resolve(line unknown, pc 0x8521eb3)
	at sun/tools/java/Environment.resolve(line unknown, pc 0x85205a2)
	at sun/tools/java/Environment.resolveName(line unknown, pc 0x851dea4)
	at sun/tools/java/Environment.resolveName(line unknown, pc 0x851dcca)
	at sun/tools/javac/SourceClass.resolveSuper(line unknown, pc 0x851ecd4)
	at sun/tools/javac/SourceClass.resolveSupers(line unknown, pc 0x852e895)
	at sun/tools/javac/SourceClass.resolveTypeStructure(line unknown, pc 0x852e1bd)
	at sun/tools/javac/SourceClass.basicCheck(line unknown, pc 0x851f971)
	at sun/tools/java/ClassDeclaration.getClassDefinition(line unknown, pc 0x851eb11)
	at sun/tools/javac/Main.compile(line unknown, pc 0x81b9992)
	at sun/tools/javac/Main.main(line unknown, pc 0x81a1d83)
error: An exception has occurred in the compiler; please file a bug report (http://java.sun.com/cgi-bin/bugreport.cgi).
1 error

so clearly something is wrong with my setup.  I'm sure I'm missing
something obvious.

I tried 'make test', but each test failed with what appears to be
essentially the same error as the one quoted above.

Thanks in advance,

-- 
Michael D, Hirsch			Work: (404) 727-7940 
Emory University, Atlanta, GA 30322	FAX: (404) 727-5611
email:  hirsch@mathcs.emory.edu		http://www.mathcs.emory.edu/~hirsch/

Public key for encrypted mail available upon request (or finger
hirsch@cssun.mathcs.emory.edu).

From kaffe@kaffe.org  Sun Oct  5 06:57:09 1997
From: kaffe@kaffe.org (Henrik Martin)
Date: Sat, 4 Oct 1997 22:57:09 -0700 (PDT)
Subject: Working version of kaffe for BSDI 3.0?
Message-ID: <199710050557.WAA29595@u2.netgate.net>


Hi folks! I'm trying to build kaffe v0.9.1 on a x86 machine running
BSDI 3.0, but it won't build. After having changed the "global" to "globl"
in the config/i386/trampolines.c file, I get the following linking error:

make -k
rm -f libkaffe_vm.a
ln -s libkaffe_vm.a.0.91 libkaffe_vm.a
gcc -g -O  -I. -I./../kaffevm -I../../config -I./../../config -I../../include -I./../../include -o kaffe main.o -L../kaffevm -lkaffe_vm  -L../../packages/tjwassoc.co.uk/APIcore/lib -lkaffe_native -L../../packages/tjwassoc.co.uk/APInet/lib -lkaffe_net -L../../packages/tjwassoc.co.uk/APIzip/lib -lkaffe_zip  -lipc -L/usr/X11R6/lib -lX11 -lkaffe_vm -lipc -lm -lc
ld: premature end of file in symbols of ../kaffevm/libkaffe_vm.a(machine.o)
*** Error code 1 (continuing)
`all' not remade because of errors.

I do have the kaffe-0.9.1-package-javasoft.com.tgz stuff installed
BTW.  Any hints on what may be wrong? If anyone has built any recent
version of kaffe successfully on BSDI 3.0, please let me know.

If you reply to the mailing list, please also cc: henrik@netgate.net.

Thanks,

/Henrik

From kaffe-announce@kaffe.org  Sun Oct  5 15:22:07 1997
From: kaffe-announce@kaffe.org (Tim Wilkinson)
Date: Sun, 05 Oct 1997 15:22:07 +0100
Subject: [FINALLY] Kaffe 0.9.2 released.
Message-ID: <3437A28F.41C67EA6@tjwassoc.co.uk>

KAFFE v0.9.2 - A JIT and interpreting virtual machine to run Java(tm)*
code
===========================================================================

                *** Kaffe is now JDK 1.1.2 compliant ***

Kaffe is a virtual machine design to execute Java bytecode.
This machine can be configured in two modes.  In one mode it operates as
a pure bytecode interpreter (not unlike Javasoft's machine); in the
second
mode if performs "just-in-time" code conversion from the abstract code
to
the host machine's native code.  This will ultimately allow execution of
Java code at the same speed as standard compiled code but while
maintaining
the advantages and flexibility of code independence.

New Features
------------

This version of Kaffe has the following new features:

 * Kaffe is now JDK 1.1.3 compliant !

 * Improved memory efficiency when allocating object.

 * Improvements to GC.
 
 * Alpha port partially fixed ... still lots of problems though. 
 
 * Lots of bug fixes (see Changelog)

Platforms
---------

Kaffe will run on the following platforms ('J' indicates 'just-in-time'
support, 'I' indicates 'interpreter' support):

 
========================================================================
  uProc     | Mode |    Systems
 
========================================================================
  i386      | J+I  |    FreeBSD 2.x +   Unixware        NetBSD 1.x
            |      |    Solaris 2.x     BSDI 2.x        Linux 2.0.0 +
            |      |    SCO 3.2v5       NeXTStep 3.x    Windows '95
            |      |    DG/UX           OpenBSD 2.x     OpenStep 4.x
            |      |    QNX 4.x
 
------------------------------------------------------------------------
  Sparc     | J+I  |    SunOS 4.x       Solaris 2.x     NetBSD 1.x
            |      |    NeXTStep 3.x    Linux           Fujitsu UXP/DS
            |      |    BSDI 3.x
 
------------------------------------------------------------------------
  Alpha     | J+I  |    Linux           OSF/1 
 
------------------------------------------------------------------------
  M68K      | J+I  |    AmigaOS         NeXTStep 3.x    NetBSD 1.x
            |      |    SunOS 4.x       Linux           AUX
            |      |    OpenBSD 2.x
 
------------------------------------------------------------------------
  PowerPC   | I    |    MkLinux         MachTen 4.0.3   AIX
 
------------------------------------------------------------------------
  MIPS      | I    |    IRIX 5 & 6      NetBSD 1.x
 
------------------------------------------------------------------------
  PARISC    | I    |    HPUX 10.x
 
------------------------------------------------------------------------  
  StrongARM | I    |    RiscIX          (work-in-progress)
 
========================================================================

I encourage people to make ports to other platforms and to feed them
back
to me.  I would recommend that initial ports concentrate on the
interpreter.
These can then be used as the basis for developing JIT versions for
these
systems.

The sources for Kaffe can be found at the following location:

Europe:
        ftp://ftp.kaffe.org/pub/kaffe/
        ftp://ftp.sarc.city.ac.uk/pub/kaffe/
        ftp://ftp.doc.ic.ac.uk/packages/kaffe/
        ftp://ftp.lysator.liu.se/pub/languages/java/kaffe/
        ftp://sunsite.mff.cuni.cz/Languages/Java/kaffe/
        ftp://sunsite.auc.dk/pub/languages/java/kaffe/
        ftp://alterego.stack.urc.tue.nl/pub/kaffe/
        ftp://ftp.fh-wolfenbuettel.de/pub/lang/java/kaffe/
        ftp://ftp.uni-erlangen.de/pub/source/Java/kaffe/
        ftp://gd.tuwien.ac.at/languages/java/kaffe/
USA:
        ftp://ftp.cs.columbia.edu/pub/kaffe/
        ftp://sunsite.unc.edu/pub/languages/java/kaffe/
Japan:
        ftp://ftp.webcity.co.jp/pub/andoh/java/kaffe/
Others:
        ftp://ftp.unicamp.br/pub/languages/java/kaffe/


What you need
-------------

For a Kaffe system you need the following parts:

 kaffe-0.9.2.tgz                        Virtual machine and all bits
                                        provided by tjwassoc.co.uk
 kaffe-0.9.2-package-javasoft.com.tgz   Sun's class libraries.

These two are are the minimum to get something useful working (you might
already have the classes libraries from JDK 1.1.2; if so don't bother to
get them again).

If you want to use graphics you'll need one of the following:

 kaffe-0.9.2-package-biss-net.com.tgz   AWT graphics support.

Uncompress and untar these archives into the same directory, configure,
make and install.

Enjoy.

Tim Wilkinson
<tim@kaffe.org>
----
* Java is a registered trademark of Sun Microsystems, Inc.

--
  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
  London, UK.                           Email:   tim@tjwassoc.co.uk

From kaffe@kaffe.org  Sun Oct  5 17:00:14 1997
From: kaffe@kaffe.org (Matthias Hopf)
Date: Sun, 5 Oct 1997 18:00:14 +0200 (MET DST)
Subject: Next Release?
In-Reply-To: <Pine.LNX.3.95.971002104801.6765B-100000@animal.blarg.net> from "James A. Hillyerd" at Oct 2, 97 10:49:10 am
Message-ID: <199710051600.SAA26186@faui01.informatik.uni-erlangen.de>

> Any guesses on when the next release of Kaffe is coming out?  This list
> has been pretty quiet lately. =)
> 
> -james

Kaffe v0.9.2 has just been released. But as I have to learn for final
exams right now, I won't be able to do anything about kaffe until
end of october. So expect a tested version of V0.9.2 (or V0.9.3 then ;)
in november.

And yes, it has been quiet here the last months. Nothing happened about
kaffe either, lots of the developers seemed to be on holiday ;)
I didn't do anything about kaffe0.9.1 as it seemed to have several
unfixed bugs.

CU

Matthias

-- 
    //             |   Matthias Hopf - "Hoeppel"   |    _      __
\\ //    Amiga     |  student of computer science  |  _|cience |-iction
 \X/ by conviction |      in Erlangen/Germany      | by belief in Future

 EMail:  mshopf@informatik.uni-erlangen.de
 Aminet: ftpamiga@epix.rrze.uni-erlangen.de
 WWW:    http://wwwcip.informatik.uni-erlangen.de/user/mshopf/

From kaffe@kaffe.org  Sun Oct  5 17:59:23 1997
From: kaffe@kaffe.org (Mark Huizer)
Date: Sun, 5 Oct 1997 18:59:23 +0200
Subject: [FINALLY] Kaffe 0.9.2 released.
In-Reply-To: <3437A28F.41C67EA6@tjwassoc.co.uk>; from "Tim Wilkinson" on Sun, Oct 05, 1997 at 03:22:07PM +0100
References: <3437A28F.41C67EA6@tjwassoc.co.uk>
Message-ID: <19971005185923.36519@xaa.stack.nl>

> 
>                 *** Kaffe is now JDK 1.1.2 compliant ***
> 
>  * Kaffe is now JDK 1.1.3 compliant !

Nice :)

What's the status of the BISS AWT classes by now? I missed out a few
weeks, so there's the lack of knowledge. Is it possible to run graphical
stuff a bit better by now?
-------------------------------------------------------------------------
- Mark Huizer  - xaa@stack.nl     - rcbamh@urc.tue.nl                   -
-------------------------------------------------------------------------
- Anyone who is capable of getting themselves made President should	-
- on no account be allowed to do the job (The Hitchhikers Guide)	-
-------------------------------------------------------------------------

From kaffe@kaffe.org  Sun Oct  5 19:53:36 1997
From: kaffe@kaffe.org (Bryan Felix)
Date: Sun, 5 Oct 1997 11:53:36 -0700 (MST)
Subject: Working version of kaffe for BSDI 3.0?
Message-ID: <199710051853.LAA20460@rapidaxcess.com>

>Hi folks! I'm trying to build kaffe v0.9.1 on a x86 machine running
>BSDI 3.0, but it won't build. After having changed the "global" to "globl"
>in the config/i386/trampolines.c file, I get the following linking error:
>
>If you reply to the mailing list, please also cc: henrik@netgate.net.
>
>Thanks,
>
>/Henrik
>
        I have done my best to get any version running on BSDI 3.0 and on
BSDI 2.1 and finally cut it off. Only so much time to spare on one project.
I decided this project is just to buggy and am waiting patiently for a
stable developement to be released.
        I did manage to get version 0.9.0 to the point that it would at
least just hang after I went through typing in the CLASSPATH by hand with
each execution. I was in touch with another user on this list that has
gotten an even older version to run on BSDI though he has to add the
CLASSPATH on to every execution as well.
        Sorry I can't be of much more help, but thought I would add my
experience for what it's worth.

With Regret,
Bryan Felix


                            _______________________________________________
        $$$$$$$$$$$$$$$$               ---Rapid Axcess---
          $$$$$$$$$$$$$$$$$    ...Premium Web Design & Domain Hosting...
           $$$$     $$$$$$$$     ______________________________________ 
           $$$$      $$$$$$$    (_____       Bryan Felix        _______)
           $$$      $$$$$   __________) felix@rapidaxcess.com  (__
          $$$     $$$$$    (_______      Voice: 602-952-8275  ______)
         $$$$  $$$$$           ____)     Page: 602-201-4646  (____________
        $$$$$$$$$             (__________ Fax: 602-224-9585 ______________)
       $$$   $$$$        $$$        
      $$$     $$$$      $$ $$       PO Box 61413 
    $$$        $$$$$   $$___$$      Phoenix, AZ 85082-1413
 $$$$$$        $$$$$$$$$     $$ ____http://www.rapidaxcess.com_____________

From kaffe@kaffe.org  Sun Oct  5 22:32:10 1997
From: kaffe@kaffe.org (Michael Hirsch)
Date: Sun, 5 Oct 1997 17:32:10 -0400 (EDT)
Subject: java.lang.ClassNotFoundException:HelloWorldApp
In-Reply-To: <199710040349.XAA22083@cssun.mathcs.emory.edu>
References: <199710040349.XAA22083@cssun.mathcs.emory.edu>
Message-ID: <199710052132.RAA23133@cssun.mathcs.emory.edu>

Okay, I found some of the problem.  I wrote:

>>>>> In article <199710040349.XAA22083@cssun.mathcs.emory.edu>, Michael Hirsch <hirsch@mathcs.emory.edu> writes:

Michael> I have the same problem that Bryan Felix reported.  I tried
Michael> to follow the README instructions exactly.  I got all four
Michael> tarballs: kaffe-0.9.1.tgz
Michael> kaffe-0.9.1-package-javasoft.com.tgz
Michael> kaffe-0.9.1-package-biss-net.com.tgz and
Michael> kaffe-0.9.1-package-epfl.ch.tgz and unpacked them in the same
Michael> directory.

Michael> ./configure worked fine, as did make and make install.

Michael> But now: 
Michael> [34] test> echo $CLASSPATH $KAFFEHOME
Michael>
.:/usr/local/share/kaffe/classes.zip:/usr/local/share/kaffe/biss.zip /usr/local/share/kaffe 
MIchael> [35] test> pwd
Michael> /usr/local/Uploads/kaffe-0.9.1/test 
Michael> [36] test> kaffe HelloWorldApp 
Michael> java.lang.ClassNotFoundException: HelloWorldApp

This, of course, does not work because the .java file has not been
compiled into a .class file.  Unfortunately, the README file says that
this should work.  Perhaps the next release will add the compiled
HelloWorldApp.class file to the test directory so silly people like me
don't get confused.

I made a compiled the .java file on another system and then tried to
run it and it worked fine.  So the interpreter/JIT seems to be okay.

Next, I tried compiling and that is still broken.  I reinstalled
everything and I still get these errors.  Any further suggestions?
It kind of looks like I don't have something installed, but I
installed all 4 .tgz files.

Thanks,

--Michael

Here are the errors:
[2] test> javac HelloWorldApp.java 
java.lang.NullPointerException
	at java/io/File.list(line unknown, pc 0x8520a6c)
	at sun/tools/java/ClassPathEntry.getFiles(line unknown, pc 0x8532197)
	at sun/tools/java/ClassPath.getFile(line unknown, pc 0x852ddb9)
	at sun/tools/java/ClassPath.getFile(line unknown, pc 0x8530bd6)
	at sun/tools/java/Package.getBinaryFile(line unknown, pc 0x853477a)
	at sun/tools/java/Package.classExists(line unknown, pc 0x852083b)
	at sun/tools/javac/BatchEnvironment.classExists(line unknown, pc 0x852c650)
	at sun/tools/java/Environment.classExists(line unknown, pc 0x8530369)
	at sun/tools/java/Environment.classExists(line unknown, pc 0x8530369)
	at sun/tools/java/Imports.importable(line unknown, pc 0x851ee55)
	at sun/tools/java/Imports.resolve(line unknown, pc 0x852d2db)
	at sun/tools/java/ImportEnvironment.resolve(line unknown, pc 0x8521eb3)
	at sun/tools/java/Environment.resolve(line unknown, pc 0x85205a2)
	at sun/tools/java/Environment.resolveName(line unknown, pc 0x851dea4)
	at sun/tools/java/Environment.resolveName(line unknown, pc 0x851dcca)
	at sun/tools/javac/SourceClass.resolveSuper(line unknown, pc 0x851ecd4)
	at sun/tools/javac/SourceClass.resolveSupers(line unknown, pc 0x852e895)
	at sun/tools/javac/SourceClass.resolveTypeStructure(line unknown, pc 0x852e1bd)
	at sun/tools/javac/SourceClass.basicCheck(line unknown, pc 0x851f971)
	at sun/tools/java/ClassDeclaration.getClassDefinition(line unknown, pc 0x851eb11)
	at sun/tools/javac/Main.compile(line unknown, pc 0x81b9992)
	at sun/tools/javac/Main.main(line unknown, pc 0x81a1d83)
error: An exception has occurred in the compiler; please file a bug report (http://java.sun.com/cgi-bin/bugreport.cgi).
1 error
[3] test> 

From kaffe@kaffe.org  Mon Oct  6 18:21:09 1997
From: kaffe@kaffe.org (P. V. Thatayya Naidu)
Date: Mon, 6 Oct 1997 10:21:09 -0700
Subject: Object Seralistion!
Message-ID: <199710061721.KAA01323@faith>

Hi,
    Will  Object Serialistion and RMI work with this new release Kaffe 0.9.2 ?
    
    
thanks
naidu
P. V. Thatayya Naidu

From kaffe@kaffe.org  Mon Oct  6 14:03:55 1997
From: kaffe@kaffe.org (Tim Wilkinson)
Date: Mon, 06 Oct 1997 14:03:55 +0100
Subject: BISS-AWT and Kaffe 0.9.2
Message-ID: <3438E1BB.1CFBAE39@tjwassoc.co.uk>

All,

Please note that I've withdrawn the BISS-AWT and EPFH.CH packages from
the latest release since neither of them work at the moment.  Hopefull
BISS-AWT will be back soon.

Regards
Tim
--
  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
  London, UK.                           Email:   tim@tjwassoc.co.uk

From kaffe@kaffe.org  Mon Oct  6 14:45:46 1997
From: kaffe@kaffe.org (Peter C. Mehlitz)
Date: Mon, 6 Oct 1997 15:45:46 +0200
Subject: [FINALLY] Kaffe 0.9.2 released.- BISS-AWT status
In-Reply-To: Mark Huizer <xaa@stack.nl>
 "Re: [FINALLY] Kaffe 0.9.2 released." (Oct  5,  6:59pm)
References: <3437A28F.41C67EA6@tjwassoc.co.uk>
 <19971005185923.36519@xaa.stack.nl>
Message-ID: <9710061545.ZM280@mwave.biss-net.com>

> Is it possible to run graphical stuff a bit better by now?

Sorry, not yet released. We have the biss.jde up ad running again, but before
we go public, this time we want to improve java.awt compatibility. This is even
more important since the java.awt classes are replaced as well. We also still
need the go from our sponsor.

Peter

From kaffe@kaffe.org  Mon Oct  6 17:25:54 1997
From: kaffe@kaffe.org (TIOOYK)
Date: Mon, 6 Oct 1997 09:25:54 -0700
Subject: Working version of kaffe for BSDI 3.0?
References: <199710051853.LAA20460@rapidaxcess.com>
Message-ID: <199710061625.JAA00292@van-as-05b01.direct.ca>

On Sun, Oct 05, 1997 at 11:53:36AM -0700, Bryan Felix wrote:
: / >Hi folks! I'm trying to build kaffe v0.9.1 on a x86 machine running
: / >BSDI 3.0, but it won't build. After having changed the "global" to "globl"
: / >in the config/i386/trampolines.c file, I get the following linking error:
: / >
: / >If you reply to the mailing list, please also cc: henrik@netgate.net.
: / >
: / >Thanks,
: / >
: / >/Henrik
: / >
: /         I have done my best to get any version running on BSDI 3.0 and on
: / BSDI 2.1 and finally cut it off. Only so much time to spare on one project.
: / I decided this project is just to buggy and am waiting patiently for a
: / stable developement to be released.
: /         I did manage to get version 0.9.0 to the point that it would at
: / least just hang after I went through typing in the CLASSPATH by hand with
: / each execution. I was in touch with another user on this list that has
: / gotten an even older version to run on BSDI though he has to add the
: / CLASSPATH on to every execution as well.
: /         Sorry I can't be of much more help, but thought I would add my
: / experience for what it's worth.
: / 

What's the problem? I use BSD/OS 2.1 and running kaffe -version I get:

# kaffe -version
Kaffe Virtual Machine
Copyright (c) 1997 T. J. Wilkinson & Associates, London, UK.
Engine: Just-in-time   Version: 0.91   Java Version: 1.1.2

I don't recall doing much to get things running, apart from the .globl
thing.

Mind you, there is so much that DOESN'T work past the tests that it's
not worth compiling right now : <

To be fair it's the graphic stuff that is pretty broken..  Perhaps
0.9.2 has a useable companion AWT library now??  Guess I'll find out
soon enough.

-g

--
Berbeee-a seys,
 Teke-a qooeeloodes in geen und gu tu a deescu reeght evey!
 Bork Bork Bork!
 Boot Kee seys, VOO-VOO!!  Nu credeet et ``Mr. Leeqoour''!!

From kaffe@kaffe.org  Mon Oct  6 23:34:36 1997
From: kaffe@kaffe.org (Tim Wilkinson)
Date: Mon, 06 Oct 1997 23:34:36 +0100
Subject: Object Seralistion!
References: <199710061721.KAA01323@faith>
Message-ID: <3439677C.3F54BC7E@tjwassoc.co.uk>

P. V. Thatayya Naidu wrote:
> 
> Hi,
>     Will  Object Serialistion and RMI work with this new release Kaffe 0.9.2 ?

I'm afraid not.  Any takers?

Cheers
Tim
--
  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
  London, UK.                           Email:   tim@tjwassoc.co.uk

From kaffe@kaffe.org  Tue Oct  7 03:22:13 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Tue, 07 Oct 1997 11:22:13 +0900
Subject: Working version of kaffe for BSDI 3.0?
In-Reply-To: Your message of Mon, 06 Oct 1997 09:25:54 MST.
Message-ID: <18855.876190933@rananim.ie.u-ryukyu.ac.jp>

Hi.

>What's the problem? I use BSD/OS 2.1 and running kaffe -version I get:

In BSD/OS 2.1 and 0.8.4, Kaffe works, but after 0.9.0, they use
stack offset in setjmp buf, which is not correnct in BSD/OS 3.0.
Ya, Bryan Felix was correct. It didn't work at all.

Here is a fix for 0.9.2

In config/i386/bsdi/md.h,  add SP_OFFSET redefine.

#include "i386/common.h"
#include "i386/threads.h"

#undef SP_OFFSET
#define SP_OFFSET               2

-- 
Shinji KONO @ Information Engineering, University of the Ryukyus
$B2OLn??<#(B @ $BN05eBg3X9)3XIt>pJs9)3X2J(B

From kaffe@kaffe.org  Tue Oct  7 07:45:29 1997
From: kaffe@kaffe.org (Orville R. Weyrich)
Date: Mon, 6 Oct 1997 23:45:29 -0700 (MST)
Subject: [FINALLY] Kaffe 0.9.2 released.
In-Reply-To: <3437A28F.41C67EA6@tjwassoc.co.uk>
Message-ID: <Pine.LNX.3.91.971006234218.28957C-100000@dopey.weyrich.com>

On Sun, 5 Oct 1997, Tim Wilkinson wrote:

>                 *** Kaffe is now JDK 1.1.2 compliant ***
 
> This version of Kaffe has the following new features:
> 
>  * Kaffe is now JDK 1.1.3 compliant !

Is one of these a typo, or is it compliant with both?

> What you need
> -------------

> 
> These two are are the minimum to get something useful working (you might
> already have the classes libraries from JDK 1.1.2; if so don't bother to
> get them again).

I have JDK 1.1.3 -- do I need to get 1.1.2, or will either work?

Thanks

orville

===================================================================
Orville R. Weyrich, Jr.                 Weyrich Computer Consulting
mailto:orville@weyrich.com                   http://www.weyrich.com
-------------------------------------------------------------------

From kaffe@kaffe.org  Tue Oct  7 13:48:38 1997
From: kaffe@kaffe.org (Gregor Hoffleit)
Date: Tue,  7 Oct 97 14:48:38 +0200
Subject: Kaffe 0.9.2 broken for NEXTSTEP
In-Reply-To: <3437A28F.41C67EA6@tjwassoc.co.uk>
References: <3437A28F.41C67EA6@tjwassoc.co.uk>
Message-ID: <9710071248.AA27266@mathi.uni-heidelberg.DE>

Kaffe 0.9.2 is broken for all NEXTSTEP architectures in various aspects, =
it won't compile (shared libs are inherently broken, trampoline code =
doesn't compile etc.pp.).


I thought I had sent Tim a NOTES/README.NEXTSTEP for inclusion in 0.9.2 =
that points out those problems and I sent in some patches, but somehow =
they didn't show up in 0.9.2.


Therefore I would opt to remove NEXTSTEP from the list of supported =
platforms for the time. Perhaps some future release can fix and =
re-install it.


For what it's worth, this is my page dedicated to kaffe on =
NEXTSTEP/OPENSTEP. You find working source and binary distributions of =
kaffe 0.8.4 and kaffe 0.9.0 for NEXTSTEP Intel, SPARC and m68k:


  http://www.mathi.uni-heidelberg.de/~flight/STEPkaffe


	Gregor



---

|
<smaller> Gregor Hoffleit                        Mathematisches =
Institut, Uni HD |

| flight@mathi.uni-heidelberg.de      INF 288, 69120 Heidelberg, Germany =
|

| (NeXTmail, MIME)                      (49)6221 54-5771    fax  54-8312 =
|

| PGP Key fingerprint =3D 23 8F B3 38 A3 39 A6 01  5B 99 91 D6 F2 AC CD =
C7 |

</smaller>

From kaffe@kaffe.org  Tue Oct  7 17:31:20 1997
From: kaffe@kaffe.org (Artur Biesiadowski)
Date: Tue, 7 Oct 1997 18:31:20 +0200 (MET DST)
Subject: Construction of primitive classes
Message-ID: <Pine.GSO.3.96L01AT.971007182047.20433A-100000@sunrise>

I'm trying to implement some of the missing native funs for
java.lang.reflect.
My question is about creating objects of one of the primitive classes -
Integer, Double, Boolean etc. 
I could created them by calling
execute_java_constructor( many args...)
but this would make quite a lot overhead - all this varargs and checks for
illegal access.
Second options is just to call  
newObject( classptr );
and the set value manually - but this could crash if class is not loaded??
Question: are primitive classes guaranteed to be loaded/compiled at this
time - calls for Field.get/Field.set are not supposed to be taken at
initialization I think.
If not, will it be enough just to check class->state and process it in
case it is not already ? 
Or just forget about speedup and stick to safe vararg constructors ?


Artur

From kaffe@kaffe.org  Tue Oct  7 17:44:08 1997
From: kaffe@kaffe.org (Vik Sohal)
Date: Tue, 07 Oct 1997 09:44:08 -0700
Subject: Object Seralistion!
Message-ID: <2.2.32.19971007164408.00c1df40@wizard.lynx.com>

Object Serialization requires an understanding of the low level protocol
that SUN uses in their own implementation. Up till a week ago this protocol
was not published, but I understand that it is now. We are looking at it and
are thinking about implementing it in Kaffe. If we do, I am quite certain we
will be able to post it as a patch to the Kaffe community.

Vik Sohal

At 11:34 PM 10/6/97 +0100, you wrote:
>P. V. Thatayya Naidu wrote:
>> 
>> Hi,
>>     Will  Object Serialistion and RMI work with this new release Kaffe
0.9.2 ?
>
>I'm afraid not.  Any takers?
>
>Cheers
>Tim
>--
>  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
>  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
>  London, UK.                           Email:   tim@tjwassoc.co.uk
>
>
>
***************************************************************
* Vik Sohal                     *  email: vik@lynx.com        *
* Lynx Real-Time Systems        *  voice: (408) 879-3900 x119 *
* 2239 Samaritan Drive          *    fax: (408) 879-3920      *
* San Jose, CA 95124            *                             *
* ------------------------------*-----------------------------*
* cute msg: "We just turned it on, and it's I.Q. is already   *
*             in the thousands!"                              *
***************************************************************

From kaffe@kaffe.org  Tue Oct  7 18:26:57 1997
From: kaffe@kaffe.org (Per Bothner)
Date: Tue, 07 Oct 1997 10:26:57 -0700
Subject: Object Seralistion!
In-Reply-To: Your message of "Tue, 07 Oct 1997 09:44:08 PDT."
 <2.2.32.19971007164408.00c1df40@wizard.lynx.com>
Message-ID: <199710071726.KAA00461@cygnus.com>

> Up till a week ago this protocol was not published,

It has been published for a while (with the Serialization "white paper"),
(though I did not realize it until someone pointed it out).

> but I understand that it is now.

Perhaps there is an improved description.

Thre are rumors that Sun is changing RMI to be more compatible with CORBA.
I don't know what this entails.  Even if they do, the old protocol will
still be useful for quite a while.

	--Per Bothner
Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner

From kaffe@kaffe.org  Tue Oct  7 23:27:30 1997
From: kaffe@kaffe.org (Alexandre Oliva)
Date: 07 Oct 1997 20:27:30 -0200
Subject: Installing kaffe on IRIX 5.2 with GNU as
Message-ID: <orafgl43al.fsf@sunsite.dcc.unicamp.br>

Hi there!

I had trouble for compiling kaffe 0.9.2 on IRIX 5.2 because GNU as was
used instead of /usr/bin/as.  The object-file threads.o produced by
GNU as wouldn't link, whereas the one produced by /usr/bin/as would.
I'd suggest that, for the particular case of IRIX, the Makefile
fragment contained a definition such as:

AS=/usr/bin/as

-- 
Alexandre Oliva
mailto:oliva@dcc.unicamp.br mailto:aoliva@acm.org
http://www.dcc.unicamp.br/~oliva
Universidade Estadual de Campinas, SP, Brasil
From kaffe@kaffe.org  Tue Oct  7 22:10:12 1997
From: kaffe@kaffe.org (Tim Wilkinson)
Date: Tue, 07 Oct 1997 22:10:12 +0100
Subject: Object Seralistion!
References: <2.2.32.19971007164408.00c1df40@wizard.lynx.com>
Message-ID: <343AA534.41C67EA6@tjwassoc.co.uk>

Vik Sohal wrote:
> 
> Object Serialization requires an understanding of the low level protocol
> that SUN uses in their own implementation. Up till a week ago this protocol
> was not published, but I understand that it is now. We are looking at it and
> are thinking about implementing it in Kaffe. If we do, I am quite certain we
> will be able to post it as a patch to the Kaffe community.

I've actually looked into this and found that the documentation for the
original JDK 1.1-beta contains more than enough information on how to
write Serialization.  You should bear in mind that most of the hard work
will be done in the Java code and only the native stubs need to be
written - and these don't worry about the format of Serialized data
anyway.  Of course, the actual function of these native methods isn't
documented so you have to guess ...  But then, thats how the rest of the
native stuff was written.

Cheers
Tim
--
  Tim Wilkinson                         Tel/Fax: +44 181 446 1309
  T. J. Wilkinson & Associates,         Mobile:  +44 370 621006
  London, UK.                           Email:   tim@tjwassoc.co.uk

From kaffe@kaffe.org  Tue Oct  7 11:04:03 1997
From: kaffe@kaffe.org (Ralf W. Stephan)
Date: Tue, 7 Oct 1997 11:04:03 +0100
Subject: Object Seralistion!
In-Reply-To: <3439677C.3F54BC7E@tjwassoc.co.uk>; from Tim Wilkinson on Mon, Oct 06, 1997 at 11:34:36PM +0100
References: <199710061721.KAA01323@faith> <3439677C.3F54BC7E@tjwassoc.co.uk>
Message-ID: <19971007110403.51428@ark.franken.de>

Tim wrote 
> P. V. Thatayya Naidu wrote:
> >     Will  Object Serialistion and RMI work with this new release Kaffe 0.9.2 ?
> I'm afraid not.  Any takers?

No, I won't bite yet, I'm still trying to understand the source.
Maybe someone can help me:

- how do you know which native functions are needed by Java's classes?
  do you just wait for LinkErrors or is there something like an API?
- many files in packages/tjwassoc... are machine generated --- from
  what input?
- can the JITC in principle take a bytecode call to a member func
  and instead direct it to a native implementaton of the class?
  Example: java.util.Vector: could this be speeded up by *directly*
  calling my own C implementation (possibly without bounds checking, e.g.)?
- is there a point in rewriting some classes to get better performance /
  free versions of them, or is this outside the kaffe project?

Thanks,
ralf
-- 
URL: http://rws.home.pages.de/

From kaffe@kaffe.org  Wed Oct  8 20:51:45 1997
From: kaffe@kaffe.org (Darrell Gallion)
Date: Wed, 08 Oct 1997 15:51:45 -0400
Subject: Configure for QNX ??
Message-ID: <343BE451.B0D6847F@cldx.com>

After running configure, make fails.
Memory model and size of int, long, .. are wrong. And as far as I know
the Watcom compiler has no 64 bit
longs.

Thanx in advance for any insight.

Darrell Gallion
dgallion@cldx.com


From kaffe@kaffe.org  Thu Oct  9 00:02:56 1997
From: kaffe@kaffe.org (Stewart Allen)
Date: Wed, 8 Oct 1997 19:02:56 -0400 (EDT)
Subject: cannot build 0.92 on linux 2.0.30
Message-ID: <Pine.LNX.3.95.971008185534.24163A-100000@flex.neuron.com>


 About 3 seconds into the build I get:


make[4]: Entering directory
`/local/hd1/home/stewart/build/kaffe-0.9.2/packages/tjwassoc.co.uk/APIcore/lib'
gcc -g -O2 -I. -I../../../../include -I./../../../../include
-I../../../../config -I./../../../../config -c -fPIC ./java.lang/Math.c
In file included from ./java.lang/Math.c:16:
../../../../config/config-math.h:24: #error "Need some form of remainder"
make[4]: *** [Math.o] Error 1
make[4]: Leaving directory
`/local/hd1/home/stewart/build/kaffe-0.9.2/packages/tjwassoc.co.uk/APIcore/lib'


 Which I traced down to the fact that Configure incorrectly determines
 that the system doesn't support "fmod" or "drem" (both of which are
 present). I can correct these settings in config.h at which point the
 entire thing builds until the kaffevm link which gives:


gcc -g -O2  -I. -I./../kaffevm -I../../config -I./../../config
-I../../include -I./../../include -o kaffe main.o -L../kaffevm -lkaffevm
../kaffevm/libkaffevm.so: undefined reference to `remainder'
make[2]: *** [kaffe] Error 1


 Any suggestions?

 -stewart-

From kaffe@kaffe.org  Thu Oct  9 02:56:24 1997
From: kaffe@kaffe.org (Glynn Clements)
Date: Thu, 9 Oct 1997 02:56:24 +0100
Subject: cannot build 0.92 on linux 2.0.30
In-Reply-To: <Pine.LNX.3.95.971008185534.24163A-100000@flex.neuron.com>
References: <Pine.LNX.3.95.971008185534.24163A-100000@flex.neuron.com>
Message-ID: <199710090156.CAA08267@cerise.sensei.co.uk>


Stewart Allen wrote:

>  About 3 seconds into the build I get:
> 
> 
> make[4]: Entering directory
> `/local/hd1/home/stewart/build/kaffe-0.9.2/packages/tjwassoc.co.uk/APIcore/lib'
> gcc -g -O2 -I. -I../../../../include -I./../../../../include
> -I../../../../config -I./../../../../config -c -fPIC ./java.lang/Math.c
> In file included from ./java.lang/Math.c:16:
> .../../../../config/config-math.h:24: #error "Need some form of remainder"
> make[4]: *** [Math.o] Error 1
> make[4]: Leaving directory
> `/local/hd1/home/stewart/build/kaffe-0.9.2/packages/tjwassoc.co.uk/APIcore/lib'
> 
> 
>  Which I traced down to the fact that Configure incorrectly determines
>  that the system doesn't support "fmod" or "drem" (both of which are
>  present). I can correct these settings in config.h at which point the
>  entire thing builds until the kaffevm link which gives:
> 
> 
> gcc -g -O2  -I. -I./../kaffevm -I../../config -I./../../config
> -I../../include -I./../../include -o kaffe main.o -L../kaffevm -lkaffevm
> .../kaffevm/libkaffevm.so: undefined reference to `remainder'
> make[2]: *** [kaffe] Error 1

This indicates that remainder isn't being replaced with fmod.

Make sure that HAVE_REMAINDER is undefined and HAVE_FMOD is defined,
e.g. insert

	#undef HAVE_REMAINDER
	#define HAVE_FMOD 1

at the top of config-math.h.

I'm running Linux 2.0.30 with libc 5.4.23, and it compiled out of the
box for me.

-- 
Glynn Clements <glynn@sensei.co.uk>

From kaffe@kaffe.org  Thu Oct  9 02:56:24 1997
From: kaffe@kaffe.org (Glynn Clements)
Date: Thu, 9 Oct 1997 02:56:24 +0100
Subject: cannot build 0.92 on linux 2.0.30
In-Reply-To: <Pine.LNX.3.95.971008185534.24163A-100000@flex.neuron.com>
References: <Pine.LNX.3.95.971008185534.24163A-100000@flex.neuron.com>
Message-ID: <199710090156.CAA08267@cerise.sensei.co.uk>


Stewart Allen wrote:

>  About 3 seconds into the build I get:
> 
> 
> make[4]: Entering directory
> `/local/hd1/home/stewart/build/kaffe-0.9.2/packages/tjwassoc.co.uk/APIcore/lib'
> gcc -g -O2 -I. -I../../../../include -I./../../../../include
> -I../../../../config -I./../../../../config -c -fPIC ./java.lang/Math.c
> In file included from ./java.lang/Math.c:16:
> .../../../../config/config-math.h:24: #error "Need some form of remainder"
> make[4]: *** [Math.o] Error 1
> make[4]: Leaving directory
> `/local/hd1/home/stewart/build/kaffe-0.9.2/packages/tjwassoc.co.uk/APIcore/lib'
> 
> 
>  Which I traced down to the fact that Configure incorrectly determines
>  that the system doesn't support "fmod" or "drem" (both of which are
>  present). I can correct these settings in config.h at which point the
>  entire thing builds until the kaffevm link which gives:
> 
> 
> gcc -g -O2  -I. -I./../kaffevm -I../../config -I./../../config
> -I../../include -I./../../include -o kaffe main.o -L../kaffevm -lkaffevm
> .../kaffevm/libkaffevm.so: undefined reference to `remainder'
> make[2]: *** [kaffe] Error 1

This indicates that remainder isn't being replaced with fmod.

Make sure that HAVE_REMAINDER is undefined and HAVE_FMOD is defined,
e.g. insert

	#undef HAVE_REMAINDER
	#define HAVE_FMOD 1

at the top of config-math.h.

I'm running Linux 2.0.30 with libc 5.4.23, and it compiled out of the
box for me.

-- 
Glynn Clements <glynn@sensei.co.uk>
From kaffe@kaffe.org  Thu Oct  9 10:10:55 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Thu, 9 Oct 1997 10:10:55 +0100
Subject: Configure for QNX ??
In-Reply-To: <343BE451.B0D6847F@cldx.com> (message from Darrell Gallion on
 Wed, 8 Oct 1997 20:51:45 +0100)
Message-ID: <9710091010.aa11291@stlind.lint.lyngso-industri.dk>


> After running configure, make fails.
> Memory model and size of int, long, .. are wrong. And as far as I know
> the Watcom compiler has no 64 bit
> longs.

Watcom 11 has 64 bit ints BUT you must use gcc for compiling kaffe on
QNX. 

Note that kaffe-0.9.2 does not compile out of the box for QNX. 

0.9.1, however, does with a little help. 

Here is the message that I posted back in july to comp.os.qnx,
describing how to compile kaffe-0.9.1 on QNX.

---------------------------------------------------------------------

I have ported the kaffe virtual machine to QNX-4.23. 

Kaffe is a free implementation of the Java virtual machine, written by
Tim Wilkinson. Kaffe uses Just-In-Time compilation of byte-code into
machine code which makes it pretty efficient (considerably faster than
Sun's "reference" interpreter based implementation - slightly slower
than Sun's JIT based implementation).

Kaffe can be used together with Sun's class libraries version
1.1.2. The complete API is not yet supported by native methods though
- I think more or less the subset corresponding to version 1.0.2 is
supported. An important exception is that awt (i.e., graphics) is not
supported by kaffe-0.9.1. 

Gnu CC 2.7.2 is required in order to compile kaffe. gcc can be
obtained from http://www.teaser.fr/~jcmichot/gcc.

The QNX patches have already been incorporated into kaffe-0.9.1, which
can be obtained from ftp://ftp.kaffe.org/pub/kaffe/kaffe-0.9.1.tgz. 

A single bugfix is required to make it compile: in
config/i386/trampolines.c replace ".global" with ".globl"

Sun's class libraries are available from a number of places, either as
part of kaffe (kaffe-0.9.1-package-javasoft.tgz), or as part of
jdk-1.1.2 which can be downloaded from javasoft.com.


Regards, Jeppe Sommer
jso@lyngso-industri.dk

----------------------------------------------------------------------

One more note: the ld command provided with gcc will confuse kaffe's
configure script because it exits zero even when unresolved symbols
are encountered. A fixed (beta) ld can be obtained from 
http://www.teaser.fr/~jcmichot/gcc/ld. I suspect that this will be a
temporary issue. 



Jeppe

From kaffe@kaffe.org  Thu Oct  9 21:42:19 1997
From: kaffe@kaffe.org (Alain Magloire)
Date: Thu, 9 Oct 1997 16:42:19 -0400 (EDT)
Subject: {K/J}NI
In-Reply-To: <3438E1BB.1CFBAE39@tjwassoc.co.uk> from "Tim Wilkinson" at Oct 6, 97 02:03:55 pm
Message-ID: <199710092042.QAA01026@pct62.cae.ca>

Bonjour M. Wilkinson

  Is kaffe supporting JNI, if not is there plan for it ?

I see that kaffe uses the old stub interface 

Thanks


--
alain

From kaffe@kaffe.org  Fri Oct 10 20:59:05 1997
From: kaffe@kaffe.org (Amitabh Arora)
Date: Fri, 10 Oct 1997 12:59:05 -0700
Subject: kaffe port for HP-UX 9.05 OS
Message-ID: <01BCD57C.4AF62E80@amitabh@aspec.com>

Has anyone done this port?  I do not need the JIT feature.
So the port should be easy.

Thanks. -amitabh arora
From kaffe@kaffe.org  Sat Oct 11 00:34:46 1997
From: kaffe@kaffe.org (James A. Hillyerd)
Date: Fri, 10 Oct 1997 16:34:46 -0700 (PDT)
Subject: JavaLobby and JavaOS
Message-ID: <Pine.LNX.3.95.971010163139.30292A-100000@animal.blarg.net>

There is some interest on the JavaLobby to write a JavaOS.  This would
probably be based on the Mach microkernel, and perhaps some pieces of GNU
Hurd.  The thought was that Kaffe would be embedded into this.  The
working name seems to be 'Oz': as in the land where anything can happen.

We could really use some C hackers, and some Kernel hackers.  Please join
in the discussion, and you should also consider joining the Java Lobby,
it's free!

http://www.javalobby.com/

Sorry if this irritates anyone.  Nobody asked me to post this here, so
direct all flames to me. =)

-james

[  James A. Hillyerd (JH2162) - james@blarg.net - Web Developer  ]
[  http://www.blarg.net/~james/   http://www.hyperglyphics.com/  ]
[ 1024/B11C3751 CA 1C B3 A9 07 2F 57 C9  91 F4 73 F2 19 A4 C5 88 ]

From kaffe@kaffe.org  Sat Oct 11 12:42:32 1997
From: kaffe@kaffe.org (Fletcher E Kittredge)
Date: Sat, 11 Oct 1997 07:42:32 -0400
Subject: JavaLobby and JavaOS
In-Reply-To: Your message of "Fri, 10 Oct 1997 16:34:46 PDT."
 <Pine.LNX.3.95.971010163139.30292A-100000@animal.blarg.net>
Message-ID: <199710111142.HAA14774@river.gwi.net>

On Fri, 10 Oct 1997 16:34:46 -0700 (PDT)  "James A. Hillyerd" wrote:
> There is some interest on the JavaLobby to write a JavaOS.  This would
> probably be based on the Mach microkernel, and perhaps some pieces of GNU
> Hurd.  The thought was that Kaffe would be embedded into this.  The
> working name seems to be 'Oz': as in the land where anything can happen.
> 
> We could really use some C hackers, and some Kernel hackers.  Please join
> in the discussion, and you should also consider joining the Java Lobby,
> it's free!

As an ex-Mach and C hacker, let me say that there are real good
technical reasons that Mach and Hurd failed.  The micro-kernel
architecture is dog slow; adding Java to the mix is not going to speed
it up.

Before wasting good Java developer time on this, please have someone
you trust go back and read all the papers from the late 80's, early
90's which detail the significant performance problems with a
micro-kernel architecture...

regards,
fletcher

From kaffe@kaffe.org  Sat Oct 11 14:26:05 1997
From: kaffe@kaffe.org (Vijay Saraswat)
Date: Sat, 11 Oct 1997 09:26:05 -0400 (EDT)
Subject: JavaLobby and JavaOS
In-Reply-To: "James A. Hillyerd"'s message of Fri, 10 Oct 1997 16:34:46 -0700 (PDT)
Message-ID: <199710111326.JAA23400@radish.research.att.com>

Oz is the name of a programming system produced by the German AI
institute (DFKI). 

From kaffe@kaffe.org  Sat Oct 11 15:51:01 1997
From: kaffe@kaffe.org (Sean McDirmid)
Date: Sat, 11 Oct 1997 07:51:01 -0700 (PDT)
Subject: JavaLobby and JavaOS
In-Reply-To: <199710111142.HAA14774@river.gwi.net>
Message-ID: <Pine.OSF.3.96.971011074458.4584A-100000@sanjuan.cs.washington.edu>

On Sat, 11 Oct 1997, Fletcher E  Kittredge wrote:

> As an ex-Mach and C hacker, let me say that there are real good
> technical reasons that Mach and Hurd failed.  The micro-kernel
> architecture is dog slow; adding Java to the mix is not going to speed
> it up.
> 
> Before wasting good Java developer time on this, please have someone
> you trust go back and read all the papers from the late 80's, early
> 90's which detail the significant performance problems with a
> micro-kernel architecture...

Now hold on here, if I understand it, the micro-kernal architecture was
slow b/c of the way it kept its form of modularity, through runtime
barriers and abstractions.  The great thing about Java code is that it is
safe and this safety can be shown (lacking in kaffe but I've written a
verifier for kimera - kimera.cs.washington.edu).  Several good performing
(and commercial) operating systems have been "derived" from
micro-kernal's, such as NT, NextStep and perhaps the upcoming BeOS. How
about a derived micro-kernal that takes Java code, compiles it, and places
it close to the micro-kernal as a trusted extension?  Now we don't have a
micro-kernal anymore but you might be able to get the performance you are
looking for. You still get the safety and abstraction that the original
micro-kernal provided (through the virtual machine compiled to machine
code interface).

Sean McDirmid
kimera.cs.washington.edu


From kaffe@kaffe.org  Sat Oct 11 17:28:18 1997
From: kaffe@kaffe.org (Bruce R. Montague)
Date: Sat, 11 Oct 1997 09:28:18 -0700
Subject: JavaLobby and JavaOS
Message-ID: <199710111628.JAA02908@madrone.cse.ucsc.edu>

 re microkernel OS engineering/performance problems - 2 recent papers 
 that may prove of interest in this regard are:

 "Experience with the Development of a Microkernel-Based, Multiserver
  Operating System", Freeman Rawson, Proc. 6th Workshop on Hot Topics
  in Operating Systems, 5-6 May 1997 (HOTOS);
 
 "The Failure of Personalities to Generalize", Brett Fleisch, ditto...

 These are also intrinsically of interest as IBM apparently lost
 $2 _billion_ on this... Quote: "Despite the level of investment and the
 adoption of what was deemed to be the best of operating systems research
 and advanced development of the 1980s, the project, in retrospect, was
 not a technical success." (Rawson).

 Apologies if this is off subject, I'm just seconding the motion to 
 think this through carefully; forewarned is forarmed, ... investing
 the effort in things like porting interfaces, validation/compatibility
 tests, etc., might be more effective?

 Regards,
 bruce

From kaffe@kaffe.org  Sat Oct 11 17:55:45 1997
From: kaffe@kaffe.org (Markus Peter)
Date: Sat, 11 Oct 1997 18:55:45 +0200
Subject: JavaLobby and JavaOS
Message-ID: <01bcd666$849cd450$0201a8c0@warpnt.spin.de>

Ok. I guess it's also time for a statement from you.
I was one of the initiators of this project.

In the mean time, we already have found somebody,
who will create a new kernel specifically for a Java OS.
This one will be very small, as a Java based OS eliminates
a lot of needs (process switching, protected memory etc...)

As soon as we have that OS, we will need a port of Kaffe
for it - I've chosen Kaffe as JIT Compiler for this OS,
because it's free and because I once needed a JIT Compiler
for Linux, and I was very disappointed of the Blackdown port,
while Kaffe appeared very excellent to me (even at that time
- it was version 0.8.4).

Unfortunately the person who will create our OS cannot
port Kaffe himself, as he's got a Java source code license
from Sun, and we want a Java OS free from any copyrights
ASAP.

That's where we will need some help from one of you for sure.
Most of the volunteers right now are experienced Java programmers,
but they do not have enough C experience to port Kaffe or it
is too long ago they used C. So if one of you is interested in
this project and would be willing to port kaffe to that OS kernel,
as soon as it is created, please contact me.

I guess without the help of one of you this project is determined
to die.

(Hehe - please do not flame me with telling me, that I should
not try to create an OS without having a clue of it, in the first place :)

Markus Peter (Developer), SPiN GmbH
warpi@spin.de   -    http://www.spin.de
Java, Perl and Web Site Development
Unite for Java!  - http://www.javalobby.org/


From kaffe@kaffe.org  Sat Oct 11 18:25:05 1997
From: kaffe@kaffe.org (roy)
Date: Sat, 11 Oct 1997 10:25:05 -0700
Subject: JavaLobby and JavaOS
In-Reply-To: <Pine.OSF.3.96.971011074458.4584A-100000@sanjuan.cs.washing
 ton.edu>
References: <199710111142.HAA14774@river.gwi.net>
Message-ID: <3.0.3.32.19971011102505.009652e0@mail.rdc1.sfba.home.net>


A microkernel is probably the worst possible fit for a 
"pure Java OS" (something that supports only Java)
if you are going to maintain the microkernel view of
things and run Kaffe as a server or subsystem.

Due to Java's security and safety, a conventional OS
carries a lot of extra baggage that is unnecessary for
a "pure Java OS".
 
The objective of microkernels is to isolate and
optimize mechanism in a small core, while providing
for multiple policies in user space. But with a 
"pure Java OS" this is unnecessary, since all you
need to support is Java. Furthermore, with a verifier
to assist, you may not need a user/kernel distinction
any more.

Instead what one probably should do is link Kaffe directly
into the kernel of some free OS (preferably real-time),
then throw out all the unnecessary stuff (user-kernel
distinction, processes, etc...)

A number of commercial and research OS vendors have used
this approach (including Chorus and the Univ. of Utah. Except
that neither is real-time, which may be important for us.)
i.e. to turning a conventional OS into a "pure java OS"
involves mostly paring stuff away.

Come to think of it, a micro-kernel may be a good starting
point if we were to link Kaffe directly into it (instead of
making it a "single server" or a "subsystem"), then turn
the non-Kaffe portions into a "pico-kernel" :-)

It may be that single-address space OSs are the best
match for Java. Perhaps we should look at Opal. Or a
kernel-extensible OS like Spin. These would probably
a much better starting point than Mach or Hurd.

By the way, I think Next is based on Mach 2.5, which is
not a microkernel. Neither is the current release of NT.

Roy


From kaffe@kaffe.org  Sat Oct 11 19:18:15 1997
From: kaffe@kaffe.org (Markus Peter)
Date: Sat, 11 Oct 1997 20:18:15 +0200
Subject: JavaOS
Message-ID: <01bcd672$0abc9c40$0201a8c0@warpnt.spin.de>

I tried emailing this mailing list earlier, but strangely, my message
did not arrive - well - here it is again - I hope you now do not
get two copies of it.

Ok. I guess it's also time for a statement from me.
I was one of the initiators of this project.

In the mean time, we already have found somebody,
who will create a new kernel specifically for a Java OS.
This one will be very small, as a Java based OS eliminates
a lot of needs (process switching, protected memory etc...)

As soon as we have that OS, we will need a port of Kaffe
for it - I've chosen Kaffe as JIT Compiler for this OS,
because it's free and because I once needed a JIT Compiler
for Linux, and I was very disappointed of the Blackdown port,
while Kaffe appeared very excellent to me (even at that time
- it was version 0.8.4).

Unfortunately the person who will create our OS cannot
port Kaffe himself, as he's got a Java source code license
from Sun, and we want a Java OS free from any copyrights
ASAP.

That's where we will need some help from one of you for sure.
Most of the volunteers right now are experienced Java programmers,
but they do not have enough C experience to port Kaffe or it
is too long ago they used C. So if one of you is interested in
this project and would be willing to port kaffe to that OS kernel,
as soon as it is created, please contact me.
I guess without the help of one of you this project is determined
to die.

(Hehe - please do not flame me with telling me, that I should
not try to create an OS without having a clue of it, in the first place :)

And one last note to Bruce Montague:
Hurd did not fail yet - in fact the first versions of Hurd were released
some months ago.
And yes - I know, lots of people say, microkernel designs have failed,
but on the other hand there are also papers citing that they are useful.
WinNT uses the MachKernel, NextStep does, Rhapsody will do.
But that does not matter anymore - after all we now have found
a developer who will create an exclusive kernel for us.

I wanted to use a microkernel in the beginning, because most
people participating in the project at the beginning were only Java
developers but were not very experienced with C programming.
So I though we should take something existing, maybe a micro-kernel,
because it is not bloated too much with unnecessary stuff.
After all a true kernel for a Java OS need not do much.


Markus Peter (Developer), SPiN GmbH
warpi@spin.de   -    http://www.spin.de
Java, Perl and Web Site Development
Unite for Java!  - http://www.javalobby.org/

From kaffe@kaffe.org  Sun Oct 12 09:37:25 1997
From: kaffe@kaffe.org (Gabriel Sebestyen)
Date: Sun, 12 Oct 1997 10:37:25 +0200
Subject: JavaOS
References: <01bcd672$0abc9c40$0201a8c0@warpnt.spin.de>
Message-ID: <34408C45.3041166C@accentcomm.com>

Hi,

First of all, Windows NT(tm :-) ) uses an own kernel developed by DEC
not Mach. Mach is designed to be a small and independent microkernel
which can be used to build your own system on. Mach 3 finally lost the
only (BSD) UN*X dependent part now it has a BSD server instead.
Some facts about Mach:
* unified and VERY efficient message handling
* simple but efficient virtual memory handling (who likes NT's memory
management? :-) )
* supports threads
* ported to a lot of platforms but it's as independent from them as it
can be

I think Mach is a real microkernel (because it was designed to be
originally and choosen by a lot of OS's) and an excellent starting point
to develope e.g. a JavaOS. Making a new kernel from the beginning does
not make sense! Just a little performance example:
An HP UNIX emulator to Mach wrote by a team for test purposes was faster
than the original HP! It's true believe me!

Gabriel Sebestyen


From kaffe@kaffe.org  Mon Oct 13 07:57:34 1997
From: kaffe@kaffe.org (Luis L. Fernandez)
Date: Mon, 13 Oct 1997 08:57:34 +0200
Subject: JavaOS
Message-ID: <199710130657.IAA00386@diogenes.iit.upco.es>


Hi,

	
	I've listening your discussion about a new Java-OS and
micro-kernels design....

	Well, in the Computer Science Department at Utah University
you can find the answer to this discussion. They have already
developed a Java-OS using a library with provides you a toolkit to
build OS named OSkit upon and Intel ix86 architecture. I've used it
and it's really good. The OSKit code it's free and it's available in

http://www.cs.utah.edu/projects/flux/oskit/

They are kernel hackers and they have redesigned the Mach micro-kernel
to run fine (search about Fluke project).

I hope this message seems useful to somebody ;-)

Best Regards,

	Luis.

---------
Luis Lopez Fernandez
e-mail: llf@computer.org

From kaffe@kaffe.org  Mon Oct 13 12:17:31 1997
From: kaffe@kaffe.org (Vijay Saraswat)
Date: Mon, 13 Oct 1997 07:17:31 -0400 (EDT)
Subject: JavaLobby and JavaOS
In-Reply-To: Sean McDirmid's message of Sat, 11 Oct 1997 07:51:01 -0700 (PDT)
Message-ID: <199710131117.HAA03790@radish.research.att.com>

   X-Authentication-Warning: vega.soi.city.ac.uk: lp set sender to owner-kaffe@soi.city.ac.uk using -f
   From: Sean McDirmid <mcdirmid@cs.washington.edu>
   Sender: owner-kaffe@soi.city.ac.uk

   On Sat, 11 Oct 1997, Fletcher E  Kittredge wrote:

   > As an ex-Mach and C hacker, let me say that there are real good
   > technical reasons that Mach and Hurd failed.  The micro-kernel
   > architecture is dog slow; adding Java to the mix is not going to speed
   > it up.
   > 
   > Before wasting good Java developer time on this, please have someone
   > you trust go back and read all the papers from the late 80's, early
   > 90's which detail the significant performance problems with a
   > micro-kernel architecture...

   Now hold on here, if I understand it, the micro-kernal architecture was
   slow b/c of the way it kept its form of modularity, through runtime
   barriers and abstractions.  The great thing about Java code is that it is
   safe and this safety can be shown (lacking in kaffe but I've written a
   verifier for kimera - kimera.cs.washington.edu).  Several good performing
   (and commercial) operating systems have been "derived" from
   micro-kernal's, such as NT, NextStep and perhaps the upcoming BeOS. How
   about a derived micro-kernal that takes Java code, compiles it, and places
   it close to the micro-kernal as a trusted extension?  Now we don't have a
   micro-kernal anymore but you might be able to get the performance you are
   looking for. You still get the safety and abstraction that the original
   micro-kernal provided (through the virtual machine compiled to machine
   code interface).

I agree with Sean here. The notion of a strongly-typed,
dynamically linked language, like Java, makes some fresh
approaches possible. 

To go a step beyond what Sean is saying: in fact it is possible
(the "declarative bytecode verification" approach) to develop
bytecode verifiers *systematically*, using a first-principles
approach, from a constraint-based specification of the opcodes
involved. One can translate an input byte code program, P,
compositionally (byte-code for byte-code) into a constraint
program, T(P) which "obviously" states the constraints on the
run-time type-state of the Java Virtual Machine, which must hold
for program execution to be type-safe. T(P) is "loop-free" and
will take time proportional to its length, which is proportional
to the length of P, to execute. If T(P) does not deadlock or
produce "false", P is type-safe.

This approach differs slightly from the one that Sean and
Javasoft have implemented. In their cases, there are separate
*byte code verifiers*, usually largish (10-15 page) C programs,
implementing a "data-flow" analysis, documented by 10-15 pages of
informal English (I have not seen Sean's documentation yet),
about whose correctness it is hard to say anything formal. 

In any case, the bottom line for this list should be, I think,
that one can assume that the subproblem of verifying the
type-safety of JVM-like bytecode is solved, with of course, one
important caveat: native code. (A lot of the "core" classes in
Java in fact contain natively implemented methods, and one has to
use separate techniques for verifying them.) 

I would not be too surprised if this approach (Java's approach)
of doing "link-time" (i.e. first-instruction-execution-time)
type-checking actually resulted in significant performance
*improvements* over standard OS's. One thing to keep in mind is
that Java's architecture is such that the bytecode verifier can
in fact accept some input code *even if it is not type-safe in
isolation* --- as long as it is type-safe when combined with the
other classes that are already loaded. That is, there are some
possibilities of doing inter-classfile optimizations (e.g.
eliminating some classcasts generated by the compiler) in a
typesafe way, that can produce even more performance benefits
than are posible if one does "straight" Java verification, as
done currently by Sun's Java systems.

Best,
Vijay

From kaffe@kaffe.org  Mon Oct 13 12:10:10 1997
From: kaffe@kaffe.org (Roger Binns)
Date: Mon, 13 Oct 1997 11:10:10 +0000 (GMT)
Subject: JavaLobby and JavaOS
In-Reply-To: <01bcd666$849cd450$0201a8c0@warpnt.spin.de> from "Markus Peter" at Oct 11, 97 06:55:45 pm
Message-ID: <199710131110.LAA25110@avon.x.co.uk>

> In the mean time, we already have found somebody,
> who will create a new kernel specifically for a Java OS.
> This one will be very small, as a Java based OS eliminates
> a lot of needs (process switching, protected memory etc...)

One thing to be careful of is drivers.  I don't know if you plan
to support commodity PC hardware, or will have very few supported
devices.

In any case, my suggestion would be to take a freeware UNIX (eg Linux)
that has lots of device support, and slowly remove/rewrite bits
to get to your goal.  This has the advantage that you have a working system 
from day one.  

(Ironically, it would also mirror the way Linux was built from Minix!)

Roger
-- 
Roger Binns   rogerb@sco.com | The reasonable man adapts himself to the world;
Software Architect           | the unreasonable one persists in trying to
Client Integration Division  | to adapt the world to himself. Therefore all
SCO, Vision Park, Cambridge  | progress depends on the unreasonable man - GBS

From kaffe@kaffe.org  Mon Oct 13 15:12:20 1997
From: kaffe@kaffe.org (Aleksey Sudakov)
Date: Mon, 13 Oct 97 10:12:20 -0400
Subject: JavaOS
In-Reply-To: <199710130657.IAA00386@diogenes.iit.upco.es>
References: <199710130657.IAA00386@diogenes.iit.upco.es>
Message-ID: <9710131412.AA07422@rkinc.com>

: 	Well, in the Computer Science Department at Utah University
: you can find the answer to this discussion. They have already
: developed a Java-OS using a library with provides you a toolkit to
: build OS named OSkit upon and Intel ix86 architecture. I've used it
: and it's really good. The OSKit code it's free and it's available in

It's totally amazing. The only bad things are that it's not available for  
immediate download and it's not free. They say that it's free for  
non-commercial use and IMHO it will affect acceptance of this totally cool  
idea.

I wonder when free software become really free.

Regards,
Aleksey

From kaffe@kaffe.org  Mon Oct 13 15:29:50 1997
From: kaffe@kaffe.org (Vijay Saraswat)
Date: Mon, 13 Oct 1997 10:29:50 -0400 (EDT)
Subject: JavaLobby and JavaOS
In-Reply-To: "Bruce R. Montague"'s message of Sat, 11 Oct 1997 09:28:18 -0700
Message-ID: <199710131429.KAA05876@radish.research.att.com>

RE: the HOTOS talks, a URL is

  http://www.eecs.harvard.edu/hotos/scribes/sessions/implementation-1.html


   Date: Sat, 11 Oct 1997 09:28:18 -0700
   From: "Bruce R. Montague" <brucem@cse.ucsc.edu>

    re microkernel OS engineering/performance problems - 2 recent papers 
    that may prove of interest in this regard are:

    "Experience with the Development of a Microkernel-Based, Multiserver
     Operating System", Freeman Rawson, Proc. 6th Workshop on Hot Topics
     in Operating Systems, 5-6 May 1997 (HOTOS);

    "The Failure of Personalities to Generalize", Brett Fleisch, ditto...

From kaffe@kaffe.org  Mon Oct 13 16:01:38 1997
From: kaffe@kaffe.org (Thanh Ma)
Date: Mon, 13 Oct 1997 11:01:38 -0400
Subject: JavaOS
In-Reply-To: <9710131412.AA07422@rkinc.com>; from Aleksey Sudakov on Mon, Oct 13, 1997 at 10:12:20AM -0400
References: <199710130657.IAA00386@diogenes.iit.upco.es> <9710131412.AA07422@rkinc.com>
Message-ID: <19971013110137.07019@dino.encore.com>

On Mon, Oct 13, 1997 at 10:12:20AM -0400, Aleksey Sudakov wrote:
> 
> It's totally amazing. The only bad things are that it's not available for  
> immediate download and it's not free. They say that it's free for  
> non-commercial use and IMHO it will affect acceptance of this totally cool  
> idea.
> 
> I wonder when free software become really free.

SCO's commercial UnixWare and OpenServer are 'free' for non-commercial
purposes :-)

Thanh

From kaffe@kaffe.org  Mon Oct 13 16:38:00 1997
From: kaffe@kaffe.org (Patrick Logan)
Date: Mon, 13 Oct 97 08:38 PDT
Subject: JavaLobby and JavaOS
In-Reply-To: <Pine.LNX.3.95.971010163139.30292A-100000@animal.blarg.net>
References: <Pine.LNX.3.95.971010163139.30292A-100000@animal.blarg.net>
Message-ID: <m0xKmZG-000DM5C@gharlane.gemstone.com>

>>>>> "James" == James A Hillyerd <james@blarg.net> writes:

    James> There is some interest on the JavaLobby to write a JavaOS.
    James> This would probably be based on the Mach microkernel, and
    James> perhaps some pieces of GNU Hurd.

Someone from Utah already ported kaffe to their OS framework. This
framework was developed by former Mach developers to *replace* Mach
because it was too unwieldy. Check it out in the research pages at
www.cs.utah.edu, I believe.

-- 
Patrick Logan                 mailto:patrickl@gemstone.com
Voice 503-533-3365            Fax   503-629-8556
Gemstone Systems, Inc         http://www.gemstone.com

From kaffe@kaffe.org  Mon Oct 13 17:02:00 1997
From: kaffe@kaffe.org (Mark Huizer)
Date: Mon, 13 Oct 1997 18:02:00 +0200
Subject: JavaOS
In-Reply-To: <9710131412.AA07422@rkinc.com>; from "Aleksey Sudakov" on Mon, Oct 13, 1997 at 10:12:20AM -0400
References: <199710130657.IAA00386@diogenes.iit.upco.es> <9710131412.AA07422@rkinc.com>
Message-ID: <19971013180200.04495@pcnov032.win.tue.nl>

The wise Aleksey Sudakov produced the following lines:
> : 	Well, in the Computer Science Department at Utah University
> : you can find the answer to this discussion. They have already
> : developed a Java-OS using a library with provides you a toolkit to
> : build OS named OSkit upon and Intel ix86 architecture. I've used it
> : and it's really good. The OSKit code it's free and it's available in
> 
> It's totally amazing. The only bad things are that it's not available for  
> immediate download and it's not free. They say that it's free for  
> non-commercial use and IMHO it will affect acceptance of this totally cool  
> idea.
> 
Can you give me some URL or mailaddress so I can look at it? Sounds
cool, and yes, I'm not commercial planning to use it :-)

Mark

From kaffe@kaffe.org  Mon Oct 13 17:43:00 1997
From: kaffe@kaffe.org (Patrick Logan)
Date: Mon, 13 Oct 97 09:43 PDT
Subject: JavaOS
In-Reply-To: <9710131412.AA07422@rkinc.com>
References: <199710130657.IAA00386@diogenes.iit.upco.es>
 <9710131412.AA07422@rkinc.com>
Message-ID: <m0xKnaj-000DM5C@gharlane.gemstone.com>

>>>>> "Aleksey" == Aleksey Sudakov <zander@rkinc.com> writes:

    Aleksey> ...They say that it's free for non-commercial use...
    Aleksey> I wonder when free software become really free.

Why do you expect to make a profit from other people's labors? Isn't
it fair that, if you are going to charge money for something that you
didn't do, that you give a portion of that money to the ones who did
the work?

(Insert story of Henny Penny here.)

Look at it this way, giving money back to the ones who did the work
will keep *them* in business, so they can do *more* work.

Enough said in this list, please mail me personally about
disagreements.

-- 
Patrick Logan                 mailto:patrickl@gemstone.com
Voice 503-533-3365            Fax   503-629-8556
Gemstone Systems, Inc         http://www.gemstone.com

From kaffe@kaffe.org  Mon Oct 13 18:17:27 1997
From: kaffe@kaffe.org (Bartlee A. Anderson)
Date: Mon, 13 Oct 1997 12:17:27 -0500
Subject: JavaOS
References: <199710130657.IAA00386@diogenes.iit.upco.es>
Message-ID: <344257A7.1B4@switch.rockwell.com>

Luis,
I saw a reference to OSkit 5 months ago and its still not available
at the site you posted. What's the deal?
Bart.
Luis L. Fernandez wrote:
> 
> Hi,
> 
> 
>         I've listening your discussion about a new Java-OS and
> micro-kernels design....
> 
>         Well, in the Computer Science Department at Utah University
> you can find the answer to this discussion. They have already
> developed a Java-OS using a library with provides you a toolkit to
> build OS named OSkit upon and Intel ix86 architecture. I've used it
> and it's really good. The OSKit code it's free and it's available in
> 
> http://www.cs.utah.edu/projects/flux/oskit/
> 
> They are kernel hackers and they have redesigned the Mach micro-kernel
> to run fine (search about Fluke project).
> 
> I hope this message seems useful to somebody ;-)
> 
> Best Regards,
> 
>         Luis.
> 
> ---------
> Luis Lopez Fernandez
> e-mail: llf@computer.org

-- 
Bartlee A. Anderson     |  Sys. Test (ISDN-Commands-Trans Link-Tools)
Rockwell International  |                Electronic Commerce Division
300 Bauman Ct.          |                    banders@ecd.rockwell.com
MS 933-605      Opinions my own, not Rockwell's  VOICE (630) 227-8975
Wood Dale, IL 60191     |                        FAX   (630) 227-9771

From kaffe@kaffe.org  Mon Oct 13 18:24:11 1997
From: kaffe@kaffe.org (Luis L. Fernandez)
Date: Mon, 13 Oct 1997 19:24:11 +0200
Subject: JavaOS
In-Reply-To: <19971013180200.04495@pcnov032.win.tue.nl>
References: <199710130657.IAA00386@diogenes.iit.upco.es>
 <9710131412.AA07422@rkinc.com>
 <19971013180200.04495@pcnov032.win.tue.nl>
Message-ID: <199710131724.TAA02484@diogenes.iit.upco.es>

Mark Huizer writes:
 > > It's totally amazing. The only bad things are that it's not available for  
 > > immediate download and it's not free. They say that it's free for  
 > > non-commercial use and IMHO it will affect acceptance of this totally cool  
 > > idea.
 > > 
 > Can you give me some URL or mailaddress so I can look at it? Sounds
 > cool, and yes, I'm not commercial planning to use it :-)
 >

Sure, you can find the OSkit at this site:

	http://www.cs.utah.edu/projects/flux/oskit/

Enjoy!!

o-------------------------------o------------------------------------o
| Luis Lopez Fernandez          |  e-mail: luisl@iit.upco.es         |   
| Industrial Systems Division   |          llf@computer.org          |
o-------------------------------o------------------------------------o
|              Technological Research Institute                      |
|          Santa Cruz de Marcenado, 26 - Planta 1 -                  |
|	            28015 Madrid - SPAIN                             |
|      Tel: 34-1-5242800 (ext. 2706) - FAX: 34-1-5423176             |
o--------------------------------------------------------------------o

From kaffe@kaffe.org  Mon Oct 13 18:42:49 1997
From: kaffe@kaffe.org (Dan Lambright)
Date: Mon, 13 Oct 1997 13:42:49 -0400
Subject: JavaLobby and JavaOS
In-Reply-To: Your message of "Mon, 13 Oct 1997 07:17:31 EDT."
 <199710131117.HAA03790@radish.research.att.com>
Message-ID: <199710131742.NAA00507@postman.opengroup.org>

In message <199710131117.HAA03790@radish.research.att.com>Vijay Saraswat writes
>   X-Authentication-Warning: vega.soi.city.ac.uk: lp set sender to owner-kaffe
>@soi.city.ac.uk using -f
.
.
.
>I agree with Sean here. The notion of a strongly-typed,
>dynamically linked language, like Java, makes some fresh
>approaches possible. 
>
>To go a step beyond what Sean is saying: in fact it is possible
>(the "declarative bytecode verification" approach) to develop
>bytecode verifiers *systematically*, using a first-principles
>approach, from a constraint-based specification of the opcodes
>involved. One can translate an input byte code program, P,
>compositionally (byte-code for byte-code) into a constraint
>program, T(P) which "obviously" states the constraints on the
>run-time type-state of the Java Virtual Machine, which must hold
>for program execution to be type-safe. T(P) is "loop-free" and
>will take time proportional to its length, which is proportional
>to the length of P, to execute. If T(P) does not deadlock or
>produce "false", P is type-safe.

This aproach is taken by Penn State's active network's program for the PLAN 
language (which has its roots in ML). I have not seen it applied to Java by 
anyone. My group is considering it. A question I have concerns the expressive 
power a "loop free" language has to the developer. What can and cannot be 
described under such constraints? I would also be interested in what tradeoffs 
work well between load-time/verification time and expressability. How large 
can a classfile be?

>
>This approach differs slightly from the one that Sean and
>Javasoft have implemented. In their cases, there are separate
>*byte code verifiers*, usually largish (10-15 page) C programs,
>implementing a "data-flow" analysis, documented by 10-15 pages of
>informal English (I have not seen Sean's documentation yet),
>about whose correctness it is hard to say anything formal. 

I have not seen a formal desription of Java's bytecode verifier. I have 
emailed Sun and Princeton's SIP group- no reply.. anyone heard of efferts 
along these lines?

>In any case, the bottom line for this list should be, I think,
>that one can assume that the subproblem of verifying the
>type-safety of JVM-like bytecode is solved, with of course, one
>important caveat: native code. (A lot of the "core" classes in
>Java in fact contain natively implemented methods, and one has to
>use separate techniques for verifying them.) 

Dont download native code unless it is authenticated from someone you trust :-)
Software fault isolation is another avenue to investigate.

>I would not be too surprised if this approach (Java's approach)
>of doing "link-time" (i.e. first-instruction-execution-time)
>type-checking actually resulted in significant performance
>*improvements* over standard OS's. One thing to keep in mind is
>that Java's architecture is such that the bytecode verifier can
>in fact accept some input code *even if it is not type-safe in
>isolation* --- as long as it is type-safe when combined with the
>other classes that are already loaded. That is, there are some
>possibilities of doing inter-classfile optimizations (e.g.
>eliminating some classcasts generated by the compiler) in a
>typesafe way, that can produce even more performance benefits
>than are posible if one does "straight" Java verification, as
>done currently by Sun's Java systems.

How big a win would this be?

From kaffe@kaffe.org  Mon Oct 13 19:47:22 1997
From: kaffe@kaffe.org (David HM Spector)
Date: Mon, 13 Oct 1997 14:47:22 -0400
Subject: JavaOS
In-Reply-To: Your message of "Mon, 13 Oct 1997 19:24:11 +0200."
 <199710131724.TAA02484@diogenes.iit.upco.es>
Message-ID: <199710131847.OAA28502@wintermute.ny.zeitgeist.com>


From the OSKit web page:

The Current Release 

	The last release of the OS Toolkit is currently unavailable pending
	the preparation of a much enhanced, somewhat incompatible version due
	in June (Oh oh, we're into July already!). Contact Jay Lepreau if you
	have an urgent need for the previous release (but don't hold your
	breath).


I am interested in evaulating OSKit for a commercial project and
wanted to find out a ball-park for licensing fees; I had mailed and
called a number of times and got no response.  Not very encouraging.

Last week on the OSKit-user list the project coordinator Jay Lepreau
stated that they would try to put up a current snaphot after the
SOPS16 conference.

OSKit looks like a very kool piece of work, but the caveat from the
authors themselves seems to be a good idea.

As for the non-commcercial use clause, it would be interesting to see
what the real story actually is.  The OSKit docs clearly state that
their work is derived from other GPLd sources including
Linux... (note: not an opinion... just an observation...)

David

-- 
-------------------------------------------------------------------------------
David HM Spector                                         spector@zeitgeist.com
Network Design & Infrastructure Security                 voice: +1 212.579.8573
Amateur Radio: W2DHM (ex-N2BCA) (ARRL life member)       GridSquare: FN30AS
-.-. --- -. -. . -.-. -  .-- .. - ....  .- -- .- - . ..- .-.  .-. .- -.. .. ---
"New and stirring things are belittled because if they are not belittled, 
the humiliating question arises, 'Why then are you not taking part in them?'"
                                                        --H. G. Wells

From kaffe@kaffe.org  Mon Oct 13 20:55:13 1997
From: kaffe@kaffe.org (James A. Hillyerd)
Date: Mon, 13 Oct 1997 12:55:13 -0700 (PDT)
Subject: Utah OSKit/JavaOS
Message-ID: <Pine.LNX.3.95.971013125028.17003A-100000@animal.blarg.net>

Here is a link to a page that talks a little about Utah's Java Computer:

http://www.cs.utah.edu/projects/flux/oskit/javaos.html

It says it will be free for non-commercial use, and that a full release
was due out May 97.  I see that they are behind on almost all of their
deadlines.  (Must be a Microsoft train ground.)

I am curious, does Kaffe's license allow someone to port it, then sell it
for commercial use?  That doesn't seem right to me.  [Perhaps the JavaOS
just plain isn't available for commercial use, since they don't explicitly
say they are selling it.]

-james

[  James A. Hillyerd (JH2162) - james@blarg.net - Web Developer  ]
[  http://www.blarg.net/~james/   http://www.hyperglyphics.com/  ]
[ 1024/B11C3751 CA 1C B3 A9 07 2F 57 C9  91 F4 73 F2 19 A4 C5 88 ]

From kaffe@kaffe.org  Mon Oct 13 20:52:48 1997
From: kaffe@kaffe.org (Pramod B. S.)
Date: Tue, 14 Oct 1997 01:22:48 +0530 (IST)
Subject: JavaOS
In-Reply-To: <19971013180200.04495@pcnov032.win.tue.nl> from "Mark Huizer" at Oct 13, 97 06:02:00 pm
Message-ID: <9710131952.AA11370@rishi.serc.iisc.ernet.in>

> Can you give me some URL or mailaddress so I can look at it? Sounds
> cool, and yes, I'm not commercial planning to use it :-)
> 
> Mark
> 
> 

http://www.cs.utah.edu/projects/flux/
look under the section titled Software Distribution Pending.

Pramod

From kaffe@kaffe.org  Tue Oct 14 02:21:46 1997
From: kaffe@kaffe.org (Neil A. Carson)
Date: Mon, 13 Oct 1997 18:21:46 -0700
Subject: Utah OSKit/JavaOS
References: <Pine.LNX.3.95.971013125028.17003A-100000@animal.blarg.net>
Message-ID: <3442C92A.2781E494@causality.com>

James A. Hillyerd wrote:

> I am curious, does Kaffe's license allow someone to port it, then sell it
> for commercial use?  That doesn't seem right to me.  [Perhaps the JavaOS
> just plain isn't available for commercial use, since they don't explicitly
> say they are selling it.]

You'd be surprised at the number of packages that are written like this:
For example, the whole of the NetBSD OS, and anything else under the
Berkeley licence.

	Cheers,

	Neil

-- 
Neil A. Carson                         Marketing Director
E-mail: neil@causality.com             Causality Limited (London, UK)
Mobile: +44 (0)370 593183              Tel/Fax: +44 (0)181 930 7408
Sent from a 233MHz StrongARM NC!       Web: http://www.causality.com.

From kaffe@kaffe.org  Tue Oct 14 02:28:11 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Mon, 13 Oct 1997 19:28:11 -0600 (MDT)
Subject: Utah OSKit/JavaOS
In-Reply-To: <Pine.LNX.3.95.971013125028.17003A-100000@animal.blarg.net> from "James A. Hillyerd" at Oct 13, 97 12:55:13 pm
Message-ID: <199710140128.TAA09531@sal.cs.utah.edu>


 Hi,

I'm the guy from the University of Utah who did the Kaffe to OS Kit
port. Unfortunately, I can't speak for my employer (Jay Lepreau), but
the situation as I see it is as follows.

We're behind our release schedule mainly for reasons of licensing which 
had to be resolved with our funders and the university. I cannot be more
specific at this point. Jay has told us that we will have at least a 
snapshot out by the end of next week. The license will allow free use 
for (at least) non-commercial purposes. The GNU-related/derived 
parts will obviously be distributed under the GNU license. 

As for the Kaffe port, it was done using an older version of the OS Kit
and using an older version of Kaffe (0.8.1). Although I don't currently
work on JavaOS, I want to make the port publicly available. As I'm doing
work-for-hire for the university, it won't be up to me to decide under
which license this will be distributed, but it will certainly not be in 
disagreement with Kaffe's license - which is a BSD-style license.
Before I make the port available, I would like to update to 0.9.2.

It is our hope that the OS Kit and the Kaffe port will spark research
in the areas that were discussed earlier in this thread. 

	- Godmar (gback@cs.utah.edu)

From kaffe@kaffe.org  Tue Oct 14 09:00:43 1997
From: kaffe@kaffe.org (Luis L. Fernandez)
Date: Tue, 14 Oct 1997 10:00:43 +0200
Subject: JavaOS
In-Reply-To: <344257A7.1B4@switch.rockwell.com>
References: <199710130657.IAA00386@diogenes.iit.upco.es>
 <344257A7.1B4@switch.rockwell.com>
Message-ID: <199710140800.KAA04485@diogenes.iit.upco.es>

Bartlee A. Anderson writes:
 > Luis,
 > I saw a reference to OSkit 5 months ago and its still not available
 > at the site you posted. What's the deal?

You are right, but this site give you a link to the person who is
responsible for OSKit project (Jay Lepreau). If you need the last
version of this library, please contact with him
(lepreau@cs.uath.edu).

Anyway if you have an urgent need I'm going to leave the oskit-0.60
release, the last one I have used, in the site:

ftp://nautilus.gsyc.inf.uc3m.es/pub/

Regards from Spain, Bart. If you need something else please, let me
know!!.

	Luis.

P.S: We have used OSKit to develop a "new" minimal micro-kernel from
the scratch. See "http://www.iit.upco.es/~luisl/advice.html" for more
info about this one. We have reduced the development time for six
months by using this toolkit. In my point, this one is a great effort
to lighten the OS development task.


o-------------------------------o------------------------------------o
| Luis Lopez Fernandez          |  e-mail: luisl@iit.upco.es         |   
| Industrial Systems Division   |          llf@computer.org          |
|                               |  http://www.iit.upco.es/~luisl     |      
o-------------------------------o------------------------------------o
|              Technological Research Institute                      |
|          Santa Cruz de Marcenado, 26 - Planta 1 -                  |
|	            28015 Madrid - SPAIN                             |
|      Tel: 34-1-5242800 (ext. 2706) - FAX: 34-1-5423176             |
o--------------------------------------------------------------------o

From kaffe@kaffe.org  Tue Oct 14 16:44:59 1997
From: kaffe@kaffe.org (Peter Michael Bertelsen)
Date: Tue, 14 Oct 1997 17:44:59 +0200
Subject: Formal JVM specification
In-Reply-To: <199710131742.NAA00507@postman.opengroup.org> (message from Dan
 Lambright on Mon, 13 Oct 1997 13:42:49 -0400)
Message-ID: <199710141544.RAA10521@ellemose.dina.kvl.dk>

Dan Lambright wrote:
> I have not seen a formal desription of Java's bytecode verifier. I have 
> emailed Sun and Princeton's SIP group- no reply.. anyone heard of efferts 
> along these lines?

I've written a report, `Semantics of Java Byte Code', giving a formal
description of the JVM and the bytecode instructions.  The JVM is
modelled as an infinite state machine, and each byte code instruction
is described by a transition rule, describing the state change and the
conditions under which execution succeeds.  The report also lists a
number of errors and ambiguities in Sun's specification of the JVM.

This might be of interest to you, although some aspects of class file
verification are not covered, e.g. checking that the file has the
correct structure.  The report is available at:

  ftp://ftp.dina.kvl.dk/pub/Staff/Peter.Bertelsen/jvm-semantics.ps.gz

Also check out the `Defensive JVM' project at Computational Logic:

  http://www.cli.com/software/djvm/

-Peter Bertelsen

From kaffe@kaffe.org  Tue Oct 14 18:00:21 1997
From: kaffe@kaffe.org (Bartlee A. Anderson)
Date: Tue, 14 Oct 1997 12:00:21 -0500
Subject: JavaOS
References: <199710130657.IAA00386@diogenes.iit.upco.es>
 <344257A7.1B4@switch.rockwell.com> <199710140800.KAA04485@diogenes.iit.upco.es>
Message-ID: <3443A525.7EE6@switch.rockwell.com>

Getting it now,
Thanks,
	Bart

Luis L. Fernandez wrote:
> 
> Bartlee A. Anderson writes:
>  > Luis,
>  > I saw a reference to OSkit 5 months ago and its still not available
>  > at the site you posted. What's the deal?
> 
> You are right, but this site give you a link to the person who is
> responsible for OSKit project (Jay Lepreau). If you need the last
> version of this library, please contact with him
> (lepreau@cs.uath.edu).
> 
> Anyway if you have an urgent need I'm going to leave the oskit-0.60
> release, the last one I have used, in the site:
> 
> ftp://nautilus.gsyc.inf.uc3m.es/pub/
> 
> Regards from Spain, Bart. If you need something else please, let me
> know!!.
> 
>         Luis.
> 
> P.S: We have used OSKit to develop a "new" minimal micro-kernel from
> the scratch. See "http://www.iit.upco.es/~luisl/advice.html" for more
> info about this one. We have reduced the development time for six
> months by using this toolkit. In my point, this one is a great effort
> to lighten the OS development task.
> 
> o-------------------------------o------------------------------------o
> | Luis Lopez Fernandez          |  e-mail: luisl@iit.upco.es         |
> | Industrial Systems Division   |          llf@computer.org          |
> |                               |  http://www.iit.upco.es/~luisl     |
> o-------------------------------o------------------------------------o
> |              Technological Research Institute                      |
> |          Santa Cruz de Marcenado, 26 - Planta 1 -                  |
> |                   28015 Madrid - SPAIN                             |
> |      Tel: 34-1-5242800 (ext. 2706) - FAX: 34-1-5423176             |
> o--------------------------------------------------------------------o

-- 
Bartlee A. Anderson     |  Sys. Test (ISDN-Commands-Trans Link-Tools)
Rockwell International  |                Electronic Commerce Division
300 Bauman Ct.          |                    banders@ecd.rockwell.com
MS 933-605      Opinions my own, not Rockwell's  VOICE (630) 227-8975
Wood Dale, IL 60191     |                        FAX   (630) 227-9771

From kaffe@kaffe.org  Wed Oct 15 10:17:22 1997
From: kaffe@kaffe.org (Malek Shabou)
Date: Wed, 15 Oct 1997 10:17:22 +0100
Subject: Pb with kaffe on win95
Message-ID: <199710150917.KAA29987@saturne.inria.fr>

 Hello
 I've compiled kaffe on a win95 succefuly, but when i try to run HelloWorldApp an error message apers:

kaffe -v -verbosejit HelloWorldApp

Loading class 'java/lang/Throwable'.
Loading class 'java/lang/IllegalArgumentException'.
Loading class 'java/lang/RuntimeException'.
Loading class 'java/lang/IllegalThreadStateException'.
Loading class 'java/lang/InterruptedException'.
Loading class 'java/lang/NoSuchMethodError'.
Loading class 'java/lang/IncompatibleClassChangeError'.
Loading class 'java/lang/LinkageError'.
Loading class 'java/lang/Error'.
Loading class 'java/lang/SecurityManager'.
Loading class 'java/lang/StringBuffer'.
Loading class 'java/lang/ThreadDeath'.
Loading class 'java/lang/ThreadGroup'.
Loading class 'java/lang/CloneNotSupportedException'.
Loading class 'java/lang/Integer'.
Loading class 'java/lang/Number'.
Loading class 'java/io/PrintStream'.
Loading class 'java/io/FilterOutputStream'.
Loading class 'java/io/OutputStream'.
Loading class 'java/lang/NullPointerException'.
Loading class 'sun/misc/VM'.
<JIT: java/lang/ThreadGroup.add(Ljava/lang/Thread;)V time 0ms (0ms) @ 0x4b82818>


(C:\KAFFE\BIN\KAFFE.EXE 1000) In cygwin_except_handler



  can any one help me?


A+


-- 
Malek Shabou				E-mail: Malek.Shabou@inria.fr
NIC France
INRIA-Rocquencourt Domaine de Voluceau
B.P.105 78153 Le Chesnay Cedex

From kaffe@kaffe.org  Wed Oct 15 16:52:09 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Thu, 16 Oct 1997 01:52:09 +1000 (EST)
Subject: NoSuchMethodError
Message-ID: <199710151552.BAA12624@zen.quick.com.au>

I was trying to get Servlets running under appache on a SunOS system
today using kaffe as the JVM.  The same apache config etc on a Solaris
system using jdk1.1.4 worked fine so I know everything other than
kaffe is cool (I tried 0.8.4,0.9.1,0.9.2).

Anyway, whenever I try to run a servlet via the jserv_mod[ule] kaffe
just says:

NoSuchMethodError

which is obviously sub-optimal from a debugging point of view.

I've not looked myself, but would it be difficult to modify things so
that one learned which method of which class was not found?

I've not yet tried using other servlet engines with kaffe - has anyone
had success with kaffe running servlets?

--sjg

From kaffe@kaffe.org  Wed Oct 15 21:48:28 1997
From: kaffe@kaffe.org (kaffe@kaffe.org)
Date: Wed, 15 Oct 1997 14:48:28 -0600 (MDT)
Subject: NoSuchMethodError
In-Reply-To: <199710151552.BAA12624@zen.quick.com.au> from "Simon J. Gerraty" at Oct 16, 97 01:52:09 am
Message-ID: <199710152048.OAA24151@lal.cs.utah.edu>

'Simon J. Gerraty' wrote:
> I've not looked myself, but would it be difficult to modify things so
> that one learned which method of which class was not found?

I've made the changes myself, but my local copy of Kaffe is whacked in
a number of respects, and you don't want to see my diff.  The actual
changes are pretty simple.  In kaffe/kaffevm/, grep for
'NoSuchMethodError', in many of the cases (e.g., in support.c) you'll
want to change those instances to 'NoSuchMethodErrorMsg(method_name)',
where "method_name" is some C string that describes the method.
For a simple hack, just use the method_name passed into the function,
in other places you may want to construct a more informative message
(e.g., "no such static method", or something.)  

Now you just need to define NoSuchMethodErrorMsg(M), which you'll put
in errors.h:
#define NoSuchMethodErrorMsg(M)  NEW_LANG_EXCEPTION_MESSAGE(NoSuchMethodError, M)

Hope this helps more than a diff....

-Pat


From kaffe@kaffe.org  Thu Oct 16 04:51:01 1997
From: kaffe@kaffe.org (Michael Madore)
Date: Wed, 15 Oct 1997 20:51:01 -0700
Subject: no math in shared library path
Message-ID: <3.0.1.32.19971015205101.006aaf88@sunshine>

Hi,

I'm running Kaffe 0.9.2 on Debian Linux (2.0.30 kernel).  When I attempt to
use the BigInteger class in my program I get the following error:

   Java.lang.UnsatisfiedLinkError:   no math in shared library path

I'm using Oracle's thin JDBC drivers, which apparently make use of this
class.  The error occurs if I use the class in a simple program without
JDBC also.  If I run the compiled program on an NT machine, I don't have
any problems.  Is this caused by my Linux installation, or by Kaffe?

TIA

Mike Madore

From kaffe@kaffe.org  Thu Oct 16 09:04:36 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Thu, 16 Oct 1997 18:04:36 +1000 (EST)
Subject: NoSuchMethodError
References: <199710152048.OAA24151@lal.cs.utah.edu>
Message-ID: <199710160804.SAA16176@gate.quick.com.au>

>'Simon J. Gerraty' wrote:
>> I've not looked myself, but would it be difficult to modify things so
>> that one learned which method of which class was not found?

>I've made the changes myself, but my local copy of Kaffe is whacked in

Thanks, I've made the appropriate changes and re-installed kaffe
I could not of course get it ti generate that error... :-)
Might have been kaffe-0.8.4 doing that :-(, anyway the patches
seem a neat addition - though I'll need to test what the result
looks like.

The good news is that after getting no where via jserv, I ran:

java sun.servlet.http.HttpServer

with appropriate CLASSPATH (java is my java_wrapper which runs kaffe) 
and using lynx, I can get SnoopServlet running.  
So presumably the problem is related to jserv.
I'll try soem other servlet engines and see how it goes.

--sjg




-- 
Simon J. Gerraty        <sjg@quick.com.au>

#include <disclaimer>   /* imagine something _very_ witty here */

From kaffe@kaffe.org  Thu Oct 16 10:13:38 1997
From: kaffe@kaffe.org (Malek Shabou)
Date: Thu, 16 Oct 1997 10:13:38 +0100
Subject: kaffe 0.9.2 with win95
Message-ID: <199710160913.KAA04563@saturne.inria.fr>

 Hello,
 i've compiled kaffe 0.9.2 under win95 envirenment, and when i trie to execute 
HelloWorldApp or any other application i get this:
Loading class 'java/lang/Throwable'.
Loading class 'java/lang/IllegalArgumentException'.
Loading class 'java/lang/RuntimeException'.
Loading class 'java/lang/IllegalThreadStateException'.
Loading class 'java/lang/InterruptedException'.
Loading class 'java/lang/NoSuchMethodError'.
Loading class 'java/lang/IncompatibleClassChangeError'.
Loading class 'java/lang/LinkageError'.
Loading class 'java/lang/Error'.
Loading class 'java/lang/SecurityManager'.
Loading class 'java/lang/StringBuffer'.
Loading class 'java/lang/ThreadDeath'.
Loading class 'java/lang/ThreadGroup'.
Loading class 'java/lang/CloneNotSupportedException'.
Loading class 'java/lang/Integer'.
Loading class 'java/lang/Number'.
Loading class 'java/io/PrintStream'.
Loading class 'java/io/FilterOutputStream'.
Loading class 'java/io/OutputStream'.
Loading class 'java/lang/NullPointerException'.
Loading class 'sun/misc/VM'.
<JIT: java/lang/ThreadGroup.add(Ljava/lang/Thread;)V time 0ms (0ms) @ 0x4b82818>


(C:\KAFFE\BIN\KAFFE.EXE 1000) In cygwin_except_handler





can any help me,

Cordialement,

-- 
Malek Shabou				E-mail: Malek.Shabou@inria.fr
NIC France
INRIA-Rocquencourt Domaine de Voluceau
B.P.105 78153 Le Chesnay Cedex


From kaffe@kaffe.org  Thu Oct 16 18:46:21 1997
From: kaffe@kaffe.org (James Cooper)
Date: Thu, 16 Oct 1997 10:46:21 -0700 (PDT)
Subject: solaris 2.5 + socket = core dump
Message-ID: <Pine.GSO.3.95.971016104047.19310B-100000@caustik.organic.com>


Hi,

I'm trying to get Apache's JServ running under kaffe
(http://java.apache.org).  JServ is a runtime environment for servlets
that work with apache.  It doesn't use any native methods of its own.

A little info about my setup:
  Sparc 5 - Solaris 2.5
  kaffe 0.9.1
  classes.zip from the Sun JDK 1.1.2

HelloWorldApp runs fine, as does the javac wrapper, so I was psyched.  I
configured JServ to use the kaffe VM instead of the Sun VM, and fired it
up.

JServ starts up fine, and begins listening on a socket.  However, when an
incoming connection comes in it dumps core.  Here's the stack trace from
gdb:

(gdb) where
#0  0xef6f4140 in _kill ()
#1  0xef6b965c in abort ()
#2  0xef4a2a1c in java_net_PlainSocketImpl_socketGetOption (this=0x276490, 
    val=15) at ./java.net/PlainSocketImpl.c:242
#3  0x29f140 in _end ()
#4  0x2a32c4 in _end ()
#5  0x2a7164 in _end ()
#6  0x271cc4 in _end ()
#7  0xef754420 in do_execute_java_method (ee=0xef774528, obj=0x14f058, 
    method_name=0x14f058 "", signature=0xef774528 "()V", mb=0x130890, 
    isStaticCall=0) at support.c:83
#8  0xef751264 in firstStartThread () at thread.c:206
#9  0xef752c30 in reschedule () at thread-internal.c:497
#10 0xc in ?? ()


I know very little about gdb, so that didn't mean much to me, but I
thought it might be helpful to someone here.

I've noticed that other folks on the list are successfully running servers
under kaffe, so I imagine it's something funky with my setup.  If anyone
has any suggestions on how to better diagnose the cause of this problem
I'd be appreciative.

Thanks

-- James

--------------------------------------------------------------------------------
                              James Paul Cooper
                      Organic Online - pixel@organic.com
                          ......Know Future......

From kaffe@kaffe.org  Fri Oct 17 05:44:26 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Fri, 17 Oct 1997 14:44:26 +1000 (EST)
Subject: kaffe can handle servlets
Message-ID: <199710170444.OAA29667@gate.quick.com.au>

FYI, a trivial mod to apache's mod_jserv.c (replace "rb+" with "r+b")
gets servlets running on SunOS 4.1.4 with kaffe-0.9.2.

When I try this on NetBSD with kaffe-0.9.1 I just get:

[Fri Oct 17 14:20:41 1997] access to /share/java/JSDK1.0.1/servlets/DateServlet failed for xx.xx.xx.xx, reason: Premature end of script headers

and of course I've not got 0.9.2 working yet.

--sjg


From kaffe@kaffe.org  Fri Oct 17 05:41:00 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Fri, 17 Oct 1997 14:41:00 +1000 (EST)
Subject: kaffe-0.9.2: core dumps on NetBSD/i386
Message-ID: <199710170441.OAA29603@gate.quick.com.au>


kaffe-0.9.2 builds ok on NetBSD/i386 but if I try and
run java foo which should produce 
java.lang.ClassNotFoundException: foo
I get
Illegal instruction (core dumped)

The stack trace looks a bit odd, so I tried:

sjg:202$ ulimit -s 32768
sjg:203$ java foo
[wait....,wait...,....]
/share: write failed, file system is full
Illegal instruction 

I would guess that something is recursing itself to death.

--sjg

From kaffe@kaffe.org  Fri Oct 17 05:25:16 1997
From: kaffe@kaffe.org (PookMan)
Date: Fri, 17 Oct 1997 00:25:16 -0400 (EDT)
Subject: Kaffe or coffee ?
Message-ID: <Pine.SUN.3.91.971017002432.8906B-100000@lisa.cs.mcgill.ca>


 I am sorry if this sounds like a stupid question but In the README file, 
we read this :

For a Kaffe system you need the following parts:

 kaffe-0.9.1.tgz                        Virtual machine and all bits
                                        provided by tjwassoc.co.uk
 kaffe-0.9.1-package-javasoft.com.tgz   Sun's class libraries.


If you want to use graphics you'll need one of the following:

 kaffe-0.9.1-package-biss-net.com.tgz   AWT graphics support.
 kaffe-0.9.1-package-epfl.ch.tgz        Alternative graphics support.


  I simply want to know where we can get those packages ?? 
At the corresponding site ? Where exactly ? ftp address ?

 THank you for your help

From kaffe@kaffe.org  Fri Oct 17 08:43:22 1997
From: kaffe@kaffe.org (PookMan)
Date: Fri, 17 Oct 1997 03:43:22 -0400 (EDT)
Subject: Kaffe or coffee ? (fwd)
Message-ID: <Pine.SUN.3.91.971017034047.16050A-100000@lisa.cs.mcgill.ca>


 Please forget my previous question and focus on this one :)
 
I am using BSDI 2.1 and I am trying to install Kaffe

 I get the following error at the compiling

 true --def ./libkaffe_bissawt.def --output-exp lib.exp --output-lib 
libkaffe_bissawt.a --dllname libkaffe_bissawt.a
touch lib.exp
ar cr libkaffe_bissawt.a.0.87 biss_awt_kernel_NativeLib.o  color.o  
display.o  font.o  graphics.o  image.o  window.o
ar: warning: biss_awt_kernel_NativeLib.o truncated to biss_awt_kernel
ranlib libkaffe_bissawt.a.0.87
rm -f libkaffe_bissawt.a
ln -s libkaffe_bissawt.a.0.87 libkaffe_bissawt.a
gcc -g -O -DNO_SHARED_LIBRARIES -I. -I. -I./../../config -I../../config 
-I../../include -I./../../include -DTRANSLATOR -I./jit -DKVER=\"0.91\" 
-c  ./jit/funcs.c
funcs.c:8805:Unknown pseudo-op:  '.global'
funcs.c:8805:Rest of line ignored. 1st junk character valued 95 (_).
*** Error code 1

Stop.
*** Error code 1

Stop.
*** Error code 1

Stop.
 
 Does anyone has any idea on how to solve this problem ?

 I would really appreciate any tip from you ...

 thank you

From kaffe@kaffe.org  Fri Oct 17 08:40:07 1997
From: kaffe@kaffe.org (Artur Biesiadowski)
Date: Fri, 17 Oct 1997 09:40:07 +0200 (MET DST)
Subject: Java inside gcc ?
Message-ID: <Pine.GSO.3.96L01AT.971016195031.2682A-100000@sunrise>



At what stage is project of integrating java/kaffe with gcc ?

Artur

From kaffe@kaffe.org  Fri Oct 17 11:50:16 1997
From: kaffe@kaffe.org (Fathi.DJEBBARI)
Date: Fri, 17 Oct 1997 11:50:16 +0100
Subject: Kaffe or coffee ? (fwd)
Message-ID: <199710171050.LAA02589@bcv64s3e.vz.cit.alcatel.fr>


pooky wrote 
> From kaffe-approval@soi.city.ac.uk Fri Oct 17 12:42 MET 1997
> X-Authentication-Warning: vega.soi.city.ac.uk: lp set sender to owner-kaffe@soi.city.ac.uk using -f
> Date: Fri, 17 Oct 1997 03:43:22 -0400 (EDT)
> From: PookMan <pooky@cs.mcgill.ca>
> To: kaffe@kaffe.org
> Subject: Kaffe or coffee ? (fwd)
> Mime-Version: 1.0
> Sender: owner-kaffe@soi.city.ac.uk
> Reply-To: kaffe@kaffe.org
> Content-Type> : > TEXT/PLAIN> ; > charset=US-ASCII> 
> Content-Length: 1056
> 
> 
>  Please forget my previous question and focus on this one :)
>  
> I am using BSDI 2.1 and I am trying to install Kaffe
> 
>  I get the following error at the compiling
> 
>  true --def ./libkaffe_bissawt.def --output-exp lib.exp --output-lib 
> libkaffe_bissawt.a --dllname libkaffe_bissawt.a
> touch lib.exp
> ar cr libkaffe_bissawt.a.0.87 biss_awt_kernel_NativeLib.o  color.o  
> display.o  font.o  graphics.o  image.o  window.o
> ar: warning: biss_awt_kernel_NativeLib.o truncated to biss_awt_kernel
> ranlib libkaffe_bissawt.a.0.87
> rm -f libkaffe_bissawt.a
> ln -s libkaffe_bissawt.a.0.87 libkaffe_bissawt.a
> gcc -g -O -DNO_SHARED_LIBRARIES -I. -I. -I./../../config -I../../config 
> -I../../include -I./../../include -DTRANSLATOR -I./jit -DKVER=\"0.91\" 
> -c  ./jit/funcs.c
> funcs.c:8805:Unknown pseudo-op:  '.global'
> funcs.c:8805:Rest of line ignored. 1st junk character valued 95 (_).
> *** Error code 1
> 
> Stop.
> *** Error code 1
> 
> Stop.
> *** Error code 1
> 
> Stop.
>  
>  Does anyone has any idea on how to solve this problem ?
> 
>  I would really appreciate any tip from you ...
> 
>  thank you
> 
> 
I am not really sure of my answer, but it seems to me that bsdi assembler
work with globl lalel instead of global.
Bye
Fathi

From kaffe@kaffe.org  Fri Oct 17 13:54:02 1997
From: kaffe@kaffe.org (Martin Lanzen)
Date: Fri, 17 Oct 1997 14:54:02 +0200 (MET DST)
Subject: Kaffe or coffee ?
In-Reply-To: <Pine.SUN.3.91.971017002432.8906B-100000@lisa.cs.mcgill.ca>
Message-ID: <199710171254.OAA27860@puh>

On 17 Okt , PookMan wrote:
> 
>  I am sorry if this sounds like a stupid question but In the README file, 
> we read this :
> 
> For a Kaffe system you need the following parts:
> 
>  kaffe-0.9.1.tgz                        Virtual machine and all bits
>                                         provided by tjwassoc.co.uk
>  kaffe-0.9.1-package-javasoft.com.tgz   Sun's class libraries.
> 
> 
> If you want to use graphics you'll need one of the following:
> 
>  kaffe-0.9.1-package-biss-net.com.tgz   AWT graphics support.
>  kaffe-0.9.1-package-epfl.ch.tgz        Alternative graphics support.
> 
> 
>   I simply want to know where we can get those packages ?? 
> At the corresponding site ? Where exactly ? ftp address ?
> 
>  THank you for your help
> 
I found these files at:
 
http://www.oasis.leo.org/java/machines/kaffe/00-index.html
-- 
                                                           
/Martin                                        

From kaffe@kaffe.org  Fri Oct 17 18:46:19 1997
From: kaffe@kaffe.org (Vik Sohal)
Date: Fri, 17 Oct 1997 10:46:19 -0700
Subject: Location of packages for kaffe
Message-ID: <2.2.32.19971017174619.00bc7a8c@wizard.lynx.com>


You can get the BISS file from:

http://btp1da.phy.uni-bayreuth.de/~werner/FreeBSD/fbsd_pub_ftp/distfiles/

The epfl file can be gotten from:

http://gd.tuwien.ac.at/languages/java/kaffe/ghindex.shtml

Vik
***************************************************************
* Vik Sohal                     *  email: vik@lynx.com        *
* Lynx Real-Time Systems        *  voice: (408) 879-3900 x119 *
* 2239 Samaritan Drive          *    fax: (408) 879-3920      *
* San Jose, CA 95124            *                             *
* ------------------------------*-----------------------------*
* cute msg: "We just turned it on, and it's I.Q. is already   *
*             in the thousands!"                              *
***************************************************************

From kaffe@kaffe.org  Fri Oct 17 20:42:20 1997
From: kaffe@kaffe.org (Per Bothner)
Date: Fri, 17 Oct 1997 12:42:20 -0700
Subject: Java inside gcc ?
In-Reply-To: Your message of "Fri, 17 Oct 1997 09:40:07 +0200."
 <Pine.GSO.3.96L01AT.971016195031.2682A-100000@sunrise>
Message-ID: <199710171941.MAA29724@cygnus.com>

> At what stage is project of integrating java/kaffe with gcc ?

Assuming you mean:  Integrating gcc-compiled Java code with Kaffe,
then:  Fairly far.  The major outstanding issue is exception handling.

	--Per Bothner
Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner

From kaffe@kaffe.org  Fri Oct 17 23:55:20 1997
From: kaffe@kaffe.org (Per Bothner)
Date: Fri, 17 Oct 1997 15:55:20 -0700
Subject: Java inside gcc ?
In-Reply-To: Your message of "Fri, 17 Oct 1997 17:20:43 CDT."
 <199710172220.RAA16096@compound.east.sun.com>
Message-ID: <199710172254.PAA08199@cygnus.com>

Tony Kimball <alk@pobox.com> writes:

> Will it be part of egcs?

That is one possibility, but it has not been decided.
(It partly depends on the egcs release plans.)

	--Per Bothner
Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner

From kaffe@kaffe.org  Sat Oct 18 02:28:10 1997
From: kaffe@kaffe.org (David Morton)
Date: Fri, 17 Oct 1997 20:28:10 -0500 (CDT)
Subject: Java inside gcc ?
In-Reply-To: <199710171941.MAA29724@cygnus.com>
Message-ID: <XFMail.971017202810.mortonda@dgm.cadvantage.com>

I'm really getting anxious... :)

On 17-Oct-97 Per Bothner wrote:
>> At what stage is project of integrating java/kaffe with gcc ?
> 
> Assuming you mean:  Integrating gcc-compiled Java code with Kaffe,
> then:  Fairly far.  The major outstanding issue is exception handling.
> 
>       --Per Bothner
> Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner

--------------------------------------------------------
David Morton         See my webpage for my public pgp key                 
E-Mail: mortonda@esuvm.emporia.edu           
WWW: http://thinker.emporia.edu/~mortonda/home.html
Date: 17-Oct-97 Time: 20:27:39  CST -600 GMT        

From kaffe@kaffe.org  Sat Oct 18 11:35:46 1997
From: kaffe@kaffe.org (Preben Randhol)
Date: 18 Oct 1997 12:35:46 +0200
Subject: Graphics
Message-ID: <jclk9fbmk8t.fsf@iq.pvv.ntnu.no>

I've tried to run a small java-program in X Windows. It works with the
ordinary java interpreter, but when I try to run it with kaffe there is just
a lot of error-msgs. It is only a small window with a button on it. Nothing
fancy. Kaffe works great with programs that only outputs text to STDOUT
though. Is there some thing I need ? I have installed the biss and eplch packages.

Thanks in advance

:info
Kaffe 0.9.1
Platform i586 with RedHat 4.2 Linux
Kernel 2.0.30

-- 
Preben Randhol                    | Soldiers are followers to the
Tlf:   73940929    Arb: 73594076  | extreme.
Epost: randhol@pvv.org            |
http://www.pvv.org/~randhol/      |     

From kaffe@kaffe.org  Sat Oct 18 21:22:57 1997
From: kaffe@kaffe.org (Corey Minyard)
Date: 18 Oct 1997 15:22:57 -0500
Subject: Java inside gcc ?
In-Reply-To: Per Bothner's message of Fri, 17 Oct 1997 12:42:20 -0700
References: <199710171941.MAA29724@cygnus.com>
Message-ID: <m2iuuuzuqm.fsf@wf-rch.cirr.com>

Per Bothner <bothner@cygnus.com> writes:

> 
> > At what stage is project of integrating java/kaffe with gcc ?
> 
> Assuming you mean:  Integrating gcc-compiled Java code with Kaffe,
> then:  Fairly far.  The major outstanding issue is exception handling.
> 
> 	--Per Bothner
> Cygnus Solutions     bothner@cygnus.com     http://www.cygnus.com/~bothner
> 

Just a thought, you might want to talk to the GNAT people (the GNU
version of Ada supported by Ada Core Technologies, if you don't know)
about exception handling.  Ada and Java exception handling are pretty
close semantically and they might have useful stuff for you.

-- 
Corey Minyard               Internet:  minyard@acm.org
  Work: minyard@nortel.ca       UUCP:  minyard@wf-rch.cirr.com

From kaffe@kaffe.org  Mon Oct 20 07:54:35 1997
From: kaffe@kaffe.org (Simon J. Gerraty)
Date: Mon, 20 Oct 1997 16:54:35 +1000 (EST)
Subject: kaffe can handle servlets
References: <199710170444.OAA29667@gate.quick.com.au>
Message-ID: <199710200654.QAA05675@gate.quick.com.au>

>FYI, a trivial mod to apache's mod_jserv.c (replace "rb+" with "r+b")
>gets servlets running on SunOS 4.1.4 with kaffe-0.9.2.

>When I try this on NetBSD with kaffe-0.9.1 I just get:

>[Fri Oct 17 14:20:41 1997] access to /share/java/JSDK1.0.1/servlets/DateServlet failed for xx.xx.xx.xx, reason: Premature end of script headers

The solution to this was the following patch:

*** kaffe-0.9.1/packages/tjwassoc.co.uk/APInet/lib/java.net/PlainSocketImpl.c.~1~    Fri Jun 27 01:02:10 1997
--- kaffe-0.9.1/packages/tjwassoc.co.uk/APInet/lib/java.net/PlainSocketImpl.c        Sun Jul 27 10:26:50 1997
***************
*** 233,243 ****
  void
  java_net_PlainSocketImpl_socketSetOption(struct Hjava_net_PlainSocketImpl* this, jint v1, jint /* bool */ v2, struct Hjava_lang_Object* v3)
  {
!       abort();
  }
  
  jint
  java_net_PlainSocketImpl_socketGetOption(struct Hjava_net_PlainSocketImpl* this, jint val)
  {
!       abort();
  }
--- 233,245 ----
  void
  java_net_PlainSocketImpl_socketSetOption(struct Hjava_net_PlainSocketImpl* this, jint v1, jint /* bool */ v2, struct Hjava_lang_Object* v3)
  {
!       /* abort(); /* no thanks --sjg */
  }
  
  jint
  java_net_PlainSocketImpl_socketGetOption(struct Hjava_net_PlainSocketImpl* this, jint val)
  {
!       /* abort(); /* no thanks --sjg */
!       val = 0;
!       return 0;
  }

-- 
Simon J. Gerraty        <sjg@quick.com.au>

#include <disclaimer>   /* imagine something _very_ witty here */

From kaffe@kaffe.org  Mon Oct 20 10:52:59 1997
From: kaffe@kaffe.org (Iso-H)
Date: Mon, 20 Oct 1997 12:52:59 +0300 (EET DST)
Subject: Solaris 2.5 + kaffe = core dump
In-Reply-To: <Pine.GSO.3.95.971016104047.19310B-100000@caustik.organic.com>
Message-ID: <Pine.LNX.3.95q.971020124525.9063A-100000@linray.ts.ray.fi>

On Thu, 16 Oct 1997, James Cooper wrote:

> 
> Hi,
> 
> I'm trying to get Apache's JServ running under kaffe
> (http://java.apache.org).  JServ is a runtime environment for servlets
> that work with apache.  It doesn't use any native methods of its own.
> 
> A little info about my setup:
>   Sparc 5 - Solaris 2.5
>   kaffe 0.9.1
>   classes.zip from the Sun JDK 1.1.2
> 

  Did you compile kaffe as dynamic? I can't get kaffe to work
  (under Solaris) if it is compiled as dynamic; it just dumps 
  core all the time!

  But if I use configure's option "--enable-staticlib" then kaffe
  works, but "KaffeVerifyBug" test gives this:

  java.lang.UnsatisfiedLinkError: no net in shared library path
        at java/lang/Throwable.<init>(line unknown, pc 16bf2c) 
        at java/lang/Error.<init>(line unknown, pc 18b654) 
        at java/lang/LinkageError.<init>(line unknown, pc 18b5dc) 
        at java/lang/UnsatisfiedLinkError.<init>(line unknown, pc 18b564) 
        at java/lang/Runtime.loadLibrary(line unknown, pc 16fbc0) 
        at java/lang/System.loadLibrary(line unknown, pc 18adf0) 
        at java/net/InetAddress.<clinit>(line unknown, pc 19219c) 
        at java/net/Socket.<init>(97)
        at KaffeVerifyBug.doit(21)
        at KaffeVerifyBug.main(9)


Iso-H

--

From kaffe@kaffe.org  Mon Oct 20 21:22:13 1997
From: kaffe@kaffe.org (Amitabh Arora)
Date: Mon, 20 Oct 1997 13:22:13 -0700
Subject: Binaries for Sun4.1 and HP 9.05 OSs
Message-ID: <01BCDD5B.2DFF26B0@amitabh@aspec.com>

Does anyone have Kaffe running on Sun 4.1.3 and HP-UX 9.05?
If yes, then please let me know how to download the binaries?

Thanks. -amitabh arora

From kaffe@kaffe.org  Mon Oct 20 23:10:10 1997
From: kaffe@kaffe.org (Alexandre Oliva)
Date: 20 Oct 1997 20:10:10 -0200
Subject: kaffe 0.9.* on IRIX 5.2: threads.s
Message-ID: <ord8l0gk71.fsf@grupiara.dcc.unicamp.br>

--Multipart_Mon_Oct_20_20:10:09_1997-1
Content-Type: text/plain; charset=US-ASCII

Hi there!

I've been discussing with Ian Lance Taylor, GNU binutils development
team leader, why I couldn't build kaffe 0.9.[012] on IRIX 5.2 if
threads.s is compiled with GNU as.  Compiling it with IRIX's built-in
as would do fine.

We couldn't decide whose fault it was exactly, but it is possible that
it is a bug in IRIX's ld, since the same object file, produced by GNU
as 2.8.1, is accepted by IRIX 5.3's ld.

Anyway, Ian has suggested the attached patch for threads.s.  It does
not affect the outcome at all, but it provides more accurate
information about the assembly code.

Regards,

-- 
Alexandre Oliva
mailto:oliva@dcc.unicamp.br mailto:aoliva@acm.org
http://www.dcc.unicamp.br/~oliva
Universidade Estadual de Campinas, SP, Brasil

--Multipart_Mon_Oct_20_20:10:09_1997-1
Content-Type: application/octet-stream; type=patch
Content-Disposition: attachment; filename="mipskaffe.diff"
Content-Transfer-Encoding: 7bit

--- config/mips/threads.s~	Fri Oct 10 20:38:39 1997
+++ config/mips/threads.s	Fri Oct 10 20:38:16 1997
@@ -11,7 +11,7 @@
  *
  */
 
-	.globl	_threadswitch
+	.globl	_threadswitch	.text
 _threadswitch:	
 	subu	$sp,$sp,32*4
 	sw	$2,  2*4($sp)
@@ -77,7 +77,7 @@
 	addu	$sp,$sp,32*4	
 	j	$31
 
-	.globl	_threadsp	
+	.globl	_threadsp	.text
 _threadsp:
 	move	$2,$sp
 	j	$31

--Multipart_Mon_Oct_20_20:10:09_1997-1--
From kaffe@kaffe.org  Tue Oct 21 03:33:13 1997
From: kaffe@kaffe.org (Michael Gesundheit)
Date: Mon, 20 Oct 1997 19:33:13 -0700 (PDT)
Subject: Kaffe bug on PC
Message-ID: <19971021023313.12883.rocketmail@send1.rocketmail.com>


Hi There,

I was able to build kaffe under VC++ 5.0. It does not
work yet but looks promissing. I have the following
problem:

In initialization time while executing initialization
of system class the vm gets into nested calls to
itself situation. In file kaffe.def:

define_insn(INVOKESTATIC)
{
......

#if defined(TRANSLATOR) && defined(HAVE_TRAMPOLINE))
  
call_indirect_const(&METHOD_NATIVECODE(method_method());
#else
   call(tmp);
#endif

........
}

In my configuration the call(tmp) is what compiled
and, if you look at this macro it is, basicly, calling
virtualMachine() which again gets back to INVOKESTATIC
case and like that - forever until the program runs
out of stack.

Any clue????

Thanks,

--Michael




_____________________________________________________________________
Sent by RocketMail. Get your free e-mail at http://www.rocketmail.com

From kaffe@kaffe.org  Tue Oct 21 18:18:27 1997
From: kaffe@kaffe.org (George Chung)
Date: Tue, 21 Oct 1997 10:18:27 -0700
Subject: Native threads and native methods on Solaris
Message-ID: <01BCDE0A.AC65E750.gchung@openhorizon.com>

Hi,
I'm new to this mailing list and couldn't really find anything in the 
archives, so here goes...

1) From looking at config/sparc/solaris2/*.c, it appears that the 
solaris/sparc implementation of Kaffe is based on a user level threads 
package. Is this correct?

2) If native methods are supported, either through JNI or through the old 
1.02 technique, then the native method library must be linked with this 
Kaffe user level threads package. Is this correct? Do you provide jacket 
routines for the synchronous system calls such as read, write, etc?

Regards,
______________________________________________________________________
George Chung
Open Horizon, Inc.
601 Gateway Blvd.
Suite 800
South San Francisco, CA 94080
t: (650) 869-2220
f: (650) 869-2201
e: gchung@openhorizon.com
w: http://www.openhorizon.com

Find out about Ambrosia(tm), the SECURE Publish/Subscribe Event
Management System for the Internet (and 100% Pure Java Certified),
at http://www.openhorizon.com.

From kaffe@kaffe.org  Tue Oct 21 19:10:33 1997
From: kaffe@kaffe.org (David Waite)
Date: Tue, 21 Oct 1997 14:10:33 -0400
Subject: ICQ for Java success?
Message-ID: <199710211839.OAA29478@m23.rickards.leon.k12.fl.us>

Has anyone had success in getting ICQ (http://www.mirabilis.com) to run
under kaffe? I've tried to get it started several times, there is a problem
with the released binaries in the linux jdk that prevents me from using them
with my system. I haven't tried kaffe yet though

ICQ is an internet pager concept: you add users to you rlist, then the
server tells you when they are on, and lets you send messages/ files and
chat with them.

-David Waite

From kaffe@kaffe.org  Wed Oct 22 08:28:50 1997
From: kaffe@kaffe.org (Nathan Gelbard)
Date: Wed, 22 Oct 1997 00:28:50 -0700
Subject: NetBSD-m68k-0.9.1 compile problem
Message-ID: <3.0.32.19971022002850.007dee70@engr.orst.edu>

I'm running NetBSD-m68k-hp300 1.2E

gcc -g -O  -I. -I. -I./../../config -I../../config -I../../include
-I./../../inc
lude -DINTERPRETER -I./intrp -DKVER=\"0.91\" -c -fpic ./intrp/machine.c
./intrp/machine.c: In function `virtualMachine':
./intrp/machine.c:141: structure has no member named `exception_table_len'
./intrp/machine.c:152: structure has no member named `ins'
./intrp/machine.c:178: structure has no member named `code'
./intrp/machine.c:182: structure has no member named `codelen'

I just grabbed the src tonight.

What to do?

	nate



[ Nathan G