[kaffe] Quick mauve test

Mark Wielaard mark at klomp.org
Sun Apr 3 15:01:26 PDT 2005


Since kaffe 1.1.5-rc and gcj/gij 4.0-cvs should now be more or less
comparable with respect to the core classes used from GNU Classpath I
ran some quick tests against mauve (cvs) for both.

In general kaffe wins :)
26843 PASSes against 25875 for gij.
This is because kaffe does actually have more recent class libraries in
CVS at the moment. Especially the locales support helps the numbers in
favor of kaffe.

Some failures people might want to look into are reported below.
All these tests pass with gij, but FAIL with kaffe. None are probably
release stoppers, but in the long run they should be fixed.

FAIL: gnu.testlet.java.lang.reflect.Method.invoke: invoke package-private method (number 1)
FAIL: gnu.testlet.java.lang.reflect.Field.promotion: byte (number 18)
FAIL: gnu.testlet.java.lang.reflect.Field.promotion: int (number 4)
FAIL: gnu.testlet.java.lang.reflect.Field.promotion: float (number 4)
FAIL: gnu.testlet.java.lang.reflect.Field.promotion: long (number 4)
FAIL: gnu.testlet.java.lang.reflect.Field.promotion: double (number 4)
FAIL: gnu.testlet.java.io.ObjectStreamClass.Test: getSerialVersionUID (number 3)
FAIL: gnu.testlet.java.io.Utf8Encoding.mojo: Surrogate Pairs (number 1)
FAIL: gnu.testlet.java.io.Utf8Encoding.mojo: Surrogate Pairs (number 2)
FAIL: gnu.testlet.java.io.Utf8Encoding.mojo: Surrogate Pairs (number 3)
FAIL: gnu.testlet.java.io.Utf8Encoding.mojo: Surrogate Pairs (number 4)
FAIL: gnu.testlet.java.io.Utf8Encoding.mojo: Surrogate Pairs (number 5)
FAIL: gnu.testlet.java.lang.Class.ClassTest: test_forName (number 10)
(In general kaffe seems better to do better with
 gnu.testlet.java.lang.Class.ClassTest then gij, except for this one.)
FAIL: gnu.testlet.java.lang.Integer.getInteger (number 19)
FAIL: gnu.testlet.java.lang.Math.max (number 3)
FAIL: gnu.testlet.java.lang.Math.max (number 19)
FAIL: gnu.testlet.java.lang.Math.min: Small doubles (number 3)
FAIL: gnu.testlet.java.lang.Math.min: Small floats (number 3)
FAIL: gnu.testlet.java.lang.Math.sin (number 1)
FAIL: gnu.testlet.java.lang.Package.getPackage: checking package for 'java.lang' (number 1)
FAIL: gnu.testlet.java.lang.String.surrogate (number 1)
FAIL: gnu.testlet.java.lang.String.surrogate (number 2)
FAIL: gnu.testlet.java.lang.String.surrogate (number 3)
FAIL: gnu.testlet.java.lang.Thread.sleep abnormal termination 142 CRASH or TIMEOUT
(This hang disappears when run mauve is ran with -verbose...)
The gnu.testlet.java.lang.Thread.security10 test seems really bad.
20 of 20 tests failed
FAIL: gnu.testlet.java.lang.ref.WeakReference.weakref: contents of weak reference (number 1)
gnu.testlet.java.lang.reflect.ReflectAccess has lots of failures, with gij there are none.
FAIL: gnu.testlet.java.net.Socket.jdk13 abnormal termination 142 CRASH or TIMEOUT
Timeout. Caused by:
java.net.SocketException: Unknown socket option
   at gnu.java.net.PlainSocketImpl.getOption (PlainSocketImpl.java:249)
   at java.net.Socket.getKeepAlive (Socket.java:1006)
FAIL: gnu.testlet.java.nio.ByteBuffer.putDouble (number 1)
FAIL: gnu.testlet.java.security.SecureRandom.SHA1PRNG: instances generate same bytes when similarly seeded (number 1)
FAIL: gnu.testlet.java.security.SecureRandom.SHA1PRNG: instances generate same bytes when similarly seeded (number 2)

I haven't actually looked into why any of the above really fails.
But hopefully someone is intrigued by these failures and will
investigate a bit more.



Escape the Java Trap with GNU Classpath!

Join the community at http://planet.classpath.org/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : http://kaffe.org/pipermail/kaffe/attachments/20050404/555483ad/attachment-0002.pgp 

More information about the kaffe mailing list