[kaffe] RandomAccessFile causes StackOverflowError

[Dalibor Topic]

Hi Ito,

--- Ito Kazumitsu <ito.kazumitsu at hitachi-cable.co.jp>
wrote:
> 
> In message "Re: [kaffe] RandomAccessFile causes
> StackOverflowError"
>     on 03/04/02, Dalibor Topic <robilad at yahoo.com>
> writes:
> 
> > I've attached some hand written tests as a start.
> They
> > have both discovered new bugs.
> 
> Copying your test programs,  I wrote two programs
> which checkes
> circular relations between read(byte[]) and
> read(byte[], int, int)
> and between write(byte[]) and write(byte[], int,
> int).

Thanks! I'll put them into the regression testing
framework once all the issues are resolved.

> The results are shown below.  What differs between
> kaffe
> and Sun's JDK is the resuls of FileInputStream and
> FileOutputStream: kaffe fails and JDK does not.  But
> as
> for other classes, both kaffe and JDK fail.

Thanks for verifying that. Could you try to create a
fix?

> IMHO, Sun's read(byte[]) and write(byte[]) in
> InputStream
> and OutputStream seem to be a simple call to
> read(byte[], int, int)
> and write(byte[], int, int),  which is the same with
> kaffe
> and GNU classpath.

Yeah, that's what the spec says. I don't think we need
to have an extra test case for that.
 
> So, as for circular relations between read(byte[])
> and
> read(byte[], int, int) and between write(byte[]) and
> write(byte[], int, int), patches are needed only for
> FileInputStream and FileOutputStream, I think.

Fine, I think my tests from yesterday exposed a flaw
in ByteArrayInputStream, too. I'll see about patching
that  issue.

cheers,
dalibor topic

__________________________________________________
Do you Yahoo!?
Yahoo! Tax Center - File online, calculators, forms, and more
http://tax.yahoo.com