https://bugs.winehq.org/show_bug.cgi?id=44055
--- Comment #23 from Gijs Vermeulen gijsvrm@gmail.com --- Created attachment 68199 --> https://bugs.winehq.org/attachment.cgi?id=68199 patch with manifest
(In reply to Zebediah Figura from comment #22)
The page for MSDN suggests the cutoff is Windows 8 (i.e. NT 6.2), not 8.1 (i.e. NT 6.3). If true, just changing the version in that patch wouldn't be enough, since supposedly the function should return TRUE if Windows 8 is in the manifest, FALSE if not.
To test this, you'd want to create a manifest, declare *only* windows 8 compatibility (i.e. no 8.1 or 10), and check if the function returns TRUE. If the documentation is accurate, it should. Unfortunately, it also means that the implementation needs to do a bit of nontrivial parsing, but at least it would look similar to 6719bf2e03.
I tested the attached patch on the testbot. It adds a manifest for the dwmapi dll that only advertises Windows 8. The result is that the testbot VMs that returned FALSE before (as seen here https://testbot.winehq.org/JobDetails.pl?Key=78963) still return FALSE (as seen here https://testbot.winehq.org/JobDetails.pl?Key=78961).
The latter testbot run is with the attached patch. Is this convincing enough that 8.1 is the cutoff?