http://bugs.winehq.org/show_bug.cgi?id=21209
--- Comment #1 from Jörg Höhle hoehle@users.sourceforge.net 2009-12-31 16:07:39 --- Created an attachment (id=25476) --> (http://bugs.winehq.org/attachment.cgi?id=25476) +dsound,+wave,warn+heap log from Ankh prior to offending commit
The attached +dsound,+wave log from wine-0.9.43+git shows that both WARN/ERR are produced, consecutively in dsound and waveoss, exactly once (at the start of the programm; the anim plays much later). Somehow I don't believe in the heap corruption hypothesis from bug #11793 rather than in an alternate code path that is now taken but does not work that well. Very interestingly, wine-1.1.31 produces no such WARN in the log, although both primary.c and wineoss/dsrender.c still contain the warning. Yet the bug remains in present versions of wine (blocked by bug #21185 since wine-1.1.35). Does this support the alternative code path or the heap corruption hypothesis? Hmm...
BTW, the music that plays at the beginning of the intro concurrently to the voices performs without pauses.
I've tested now with and without WINEDEBUG=warn+heap in wine-1.1.31 plus 0.9.43+git upto as well as before this commit; nothing changed w.r.t. the bug symptoms (3 versions, each test repeated twice for a total of 6 runs). Sometimes heap arena errors were mentioned, sometimes not.