http://bugs.winehq.org/show_bug.cgi?id=12319
--- Comment #3 from Austin English austinenglish@gmail.com 2008-04-01 14:49:57 --- (In reply to comment #2)
Yes, I agree it's odd that a change in user32 could cause something like this to happen, however I definitely did the regression test correctly. Just in case, I also tried with a1bc05e4c0eb017eda7636343af80d18b1f64d6c (the commit right before the one mentioned by the bisect), which was fine.
Still, very strange, can you try:
A) Compiling just before the bad commit, and retest (which you did, but try with a clean ~/.wine to be sure). B) Compiling at the bad commit and retest (again, with a clean ~/.wine) C) Reverting that commit on current git, and test with a clean ~/.wine