http://bugs.winehq.org/show_bug.cgi?id=27168
--- Comment #70 from Ilya Konovalov aragaer@gmail.com 2012-09-25 08:12:05 CDT --- (In reply to comment #69)
So we need
- a test showing Windows' behavior,
- another test showing the behavior of the above patch, and
- a comparison of the results
in order to get the patch merged?
Unfortunately no, the patch above is not likely to be ever merged.
We need - a test showing the difference between windows and current crypt32/secur32 implementation (sent to wine-patches already, don't know the current state though) - some "hacky" patch "fixing" crypt32 - simple patch to secur32 (could be taken from patch above)
"Hacky" means it is not really doing complex job of keeping reference counters and stuff and instead .. doing something simple but allowing us to keep certificate storage without reference to it. Windows implementation does that (and I don't doubt it's indeed some workaround).
The problem is that we can't look into windows' solution as it would be reverse engineering and it is not something we want. The solution discussed here in #21 - #28 (before considering a more generic way) might work. Or might not.