--- Gwenole Beauchesne <[EMAIL PROTECTED]> wrote: > Hi, > > > You might get a more useful test failure if you > compile with jikes. > > Then, ideally, one of the actual tests will fail > (vs. compilation of > > the test) and that should help pinpoint the > problem. > > Guess what? I could manually compile > ThreadState.java for example. That > is: > . ../../BUILD_ENVIRONMENT > $JAVA at.dms.kjc.Main ThreadState.java > > Then running it prints three "success" then > segfaults. > > - Interestingly another test that failed (internal > error while > compiling) was URLTest. Again, manually compiling it > succeeds. Likewise > for running it. That's with the "pthreads" > implementation. I could > manually compile many of those tests that failed to > compile at make > check. > > - Case of DoubleComp.java. Even manually compiling > it fails with kjc > (internal error, java/lang/NullPoinerException > caught). However, > compiling it with jikes succeeds. The test run > succeeds too.
That sounds really weird. Just a suggestion: could you try to recompile using gcc 3.1 and confirm that the bugs still happen ? gcc 3.0.x is a rather buggy gcc in my experience. cheers, dalibor topic __________________________________________________ Do You Yahoo!? Yahoo! - Official partner of 2002 FIFA World Cup http://fifaworldcup.yahoo.com _______________________________________________ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
