Re: Java test failures

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Mark" == Mark Mitchell <mark@codesourcery.com> writes:

Mark>   FAIL: Thread_Wait execution from source compiled test
Mark>   FAIL: Thread_Wait execution from bytecode->native test
Mark>   FAIL: Thread_Wait -O execution from source compiled test

Mark> I assume this is because I did not configure with --enable-threads.

Yup.

Mark> If some programs work and some don't, that's OK -- but then
Mark> these tests should be XFAIL'd when --enable-threads is not used.

I agree.  I've written a patch to do this.  I've tested it on my posix
threads build.  I'll try to build a no-threads gcc tree tomorrow and
see if it does the right thing.

Mark> Eliminating FAILs is important because it is only when tests are
Mark> normally expected to pass that new FAILs stand out.

I agree, and we've already decided on and implemented a policy of 0
FAIL for libgcj.  We also work to reduce the XPASS count as much as
possible (sometimes it isn't easily doable to make the XPASS count
always be 0).  The nicest thing about 0 FAIL is that you immediately
know if something has gone wrong.

Tom