http://bugs.winehq.org/show_bug.cgi?id=12279
--- Comment #11 from billstei billstei@hbci.com 2008-07-06 07:56:47 --- I am far too suspicious of the bisect/regression testing at this point to post the results, they are probably bogus. To prove this I got the 0.9.50 source code directly (not using git) and compiled that. In theory my build of 0.9.50 and the Ubuntu 0.9.50 build should give me the exact same behaviours. The Ubuntu pre-built package works and mine does not. Doing a file compare the dlls msi.dll and shdocvw.dll are the main ones that differ, other than the registry itself in system.reg and user.reg. Until I get a valid 0.9.50 build I cannot expect reasonable results from the bisect because the bisect begins with the statement: git bisect good wine-0.9.50, and that is not true statement.