java/decl.c patch for clear_binding_level

Tom Tromey tromey@redhat.com
Sat Mar 17 17:01:00 GMT 2001


>>>>> "Zack" == Zack Weinberg <zackw@stanford.edu> writes:

Zack> # of unexpected failures        96

This is still way too high.  The "real" number is around 2.  It
"should" be 0 -- I went through the test suite a while back and set
things up so that the number of unexpected failures is 0.  This is
supposed make it easy to see when we introduce regressions.  The fact
that it is nonzero at all means that someone has not been paying
attention to the test suite (I'm not casting blame -- it very well
could be me).

Zack> (it uses XFAIL for execution tests when the program failed to
Zack> compile - shouldn't that be UNTESTED?)

It doesn't matter that much to me.

Zack> PASS: Thread_Join -O compilation from source
Zack> libgcj failure: only thread already running

This means you failed to configure with `--enable-threads=posix'.

Zack> FAIL: Thread_Join -O execution from source compiled test
Zack> XFAIL: Thread_Join -O execution from source compiled test
Zack> (why does that get both a FAIL and an XFAIL?)

No clue.

Zack> and a very few 'compilation from bytecode' failures with no
Zack> error message attached,

I've seen this in the logs but I have not investigated it.

Tom



More information about the Java-patches mailing list