On Sat, Jul 04, 2009 at 02:41:13PM -0400, Andrew Whitworth wrote: > > I used mem_allocate_zeroed_typed in r39881, and fixed this second > segfault in r39882. I no longer see any GC-related bugs in the rakudo > spectest suite. Until a new one is found, I'm declaring "Mission > Accomplished!".
...Are you also wearing a flight suit and standing on the deck of an aircraft carrier? :-) :-) With rakudo 6a4d66a, parrot r39883, and pugs r27409, I still see gc-related bugs in the spectest output. As before, I started with completely fresh checkouts and builds of rakudo and parrot for these tests. Legend: [*] = fails due to known rakudo bug [a] = passes when run from 'make spectest' or 'make <testfile>' [b] = passes when run from 'perl6' fakecutable [c] = passes when run via 'parrot perl6.pbc' [d] = passes when run via 'parrot -G perl6.pbc' x86 spectest results: [ d] t/spec/S02-builtin_data_types/anon_block. ?? ?? ?? [ d] t/spec/S04-declarations/constant.rakudo ?? ?? ?? [ d] t/spec/S05-transliteration/trans.rakudo 51 4 8-10 14 [ ] t/spec/S10-packages/basic.rakudo 0 6 22 0 ?? [ d] t/spec/S32-num/log.rakudo ?? ?? ?? x86_64 spectest results: [ cd] t/spec/S04-statements/return.rakudo ?? ?? ?? [ ] t/spec/S10-packages/basic.rakudo 0 6 22 0 ?? [ cd] t/spec/S32-hash/exists.rakudo ?? ?? ?? Of course, the desired result is "[abcd]" for each of the above. It's also worth noting that with the exception of the S10-packages test (which is a double-free bug in Parrot), the tests that are failing now are completely different from the ones in my original report yesterday. Lastly, I must add that I'm *really* happy about and appreciative of the excellent progress that's taken place on this issue since yesterday. But I fear we're not quite "there" yet, so please keep up the great work! Pm _______________________________________________ http://lists.parrot.org/mailman/listinfo/parrot-dev
