[PATCH] Use Windows 10 as default prefix version.

[Nikolay Sivov]

On 1/21/22 18:44, Paul Gofman wrote:
> On 1/21/22 18:32, Nikolay Sivov wrote:
>>
>>
>> On 1/21/22 18:11, Paul Gofman wrote:
>>> There is also version number in kernel32/version.rc which some games 
>>> expect to have in sync IIRC.
>>
>> Do you mean a build number? Is it matching now?
>
> No, it doesn't. It is currently from Windows 10 1909 (18362). Maybe I 
> am messing up that something known was depending on the match (maybe 
> only on that version being recent enough). But maybe it still makes 
> sense to make it consistent?

Yes, I think so. I'd like to have this done step by step so if something 
breaks we can track why.

>>
>>>
>>> Then, there is a version number hardcoded in 
>>> kernelbase/version.c:version_data[] which is currently 17134, that 
>>> should probably be bumped as well? (fwiw there is also the number in 
>>> ntdll/version.c: VersionData[] but this is 17763 now.
>>
>> That means it's already doesn't match if I switch to Win10 manually. 
>> We should probably use some header to make sure all of this is 
>> updated properly.
>
> Yes, it probably doesn't match now as well. We probably can't make 
> manually switched Win version fully consistent. But still I guess it 
> would be more straightforward if we default to recent enough Win10 
> with all the known version queries matching for the default case.
>
>
>>
>>>
>>> Probably not exactly related but maybe once the default is upgraded 
>>> to Win10 it makes sense to add ReleaseId and DisplayVersion registry 
>>> values? I recall at least one game depending in ReleaseId presence.
>>
>> Sure. Are those values new to Windows 10?
>
> Yes, I think that was introduced in Win10. And DisplayVersion is in 
> fact probably even newer, not sure it actually exists on 1809 (that 
> is, 17763). But also, maybe it would make sense to bump to something 
> newer at once? 17763 is more than 3 years old already and there are 
> apps which refuse to work or at least complain about outdated version 
> (the most fresh requirement I saw so far was 19041, 2004 from May 2020).
>
>
I don't see a problem with bumping it up to 21H1 version too. That's 
definitely a separate change.