Re: [egit-dev] JGit Build Update

[Igor Fedorenko <ifedorenko@xxxxxxxxxxxx>]

This could a bug in IBM JVM itself. Running the tests in debugger or
with "export JAVA_COMPILER=NONE" on cli makes the problem go away. At
least on x86-64 linux and using IBM 1.6 SR6.

--
Regards,
Igor

Shawn Pearce wrote:
> On Thu, Dec 10, 2009 at 4:01 PM, Matthias Sohn
> <matthias.sohn@xxxxxxxxxxxxxx> wrote:
> > Under Maven 3.0-alpha5 on Windows XP using Sun Java6
> >
> > $ java -version
> > java version "1.6.0_17"
> > Java(TM) SE Runtime Environment (build 1.6.0_17-b04)
> > Java HotSpot(TM) Client VM (build 14.3-b01, mixed mode, sharing)
> >
> > on latest jgit at e51ebb1c8 all tests pass.
>
> Thanks.
>
> Chris and I are pretty sure there is a bug in the MessageDigest
> implementation in the Java 6 version of J9... or there is a bug in the
> Sun implementation of MessageDigest which doesn't follow the spec and
> JGit is assuming the broken implementation.
>
> We're not sure which it is though.  The offending code according to J9 is doing:
>
>   byte[] buf = new byte[72];
>   MessageDigest md  ...
>   md.update(buf, 0, 42);
>
> and the md.update call throws with an IllegalArgumentException over
> one of the last 2 arguments (offset in array to read from, and number
> of bytes at that offset to read) as being invalid.  Seriously WTF.
> Either the API is horribly wrong, or J9 is busted.
> _______________________________________________
> egit-dev mailing list
> egit-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/egit-dev