http://bugs.winehq.org/show_bug.cgi?id=21071
--- Comment #12 from Saulius K. saulius2@gmail.com 2010-01-02 02:45:05 --- (in reply to Comment #5)
Looks like I screwed up STGM_SIMPLE mode. The first time StorageImpl_StreamWriteAt is used, it will assume the stream has been set to the size it asked for and try to write to a small block chain instead of a big block chain.
Vincent, is it feasible to write some test case to prevent such regression in future?