http://bugs.winehq.org/show_bug.cgi?id=3689
--- Comment #38 from Anastasius Focht focht@gmx.net 2009-04-02 18:29:42 --- Hello,
--- quote --- Thanks, Anastasius. My first question is: are these dlls the same by their exports? --- quote ---
At the time of Win95 the Operating System one and Office9x shipped one were probably similar in exports/functionality. Nowadays oleaut32 versions are much more evolved than Office9x shipped one which dates back to 1996.
--- quote --- Then Wine could just simply fake one dll with another. --- quote ---
As long as Office 97 components don't depend on internal features/quirks/behaviour of that old oleaut32.dll, yes. For tests just symlink oleaut32.dll.so to 32autole.dll.so and change the load order to builtin-then-native. A placeholder in system32 is not needed.
--- quote --- The second is: couldn't that be some other bug in Wine which prevents 32autole.dll from registering "stdole32.tlb" in v2 format correctly? Of course, I could take the TLB from Wine tree and investigate this on real Windows myself. --- quote ---
If you question my analysis you are free to investigate.
--- quote --- The third question (not so personal) would be:
Couldn't ITypeLib2_Constructor_SLTG() be used (from dlls/oleaut32/typelib.c) to generate v1 format in Wine? --- quote ---
You refer to the typelib loading part. What you need is v1 typelib writer code like Wine's IDL compiler does for v2 typelibs (write_msft.c).
Regards