std.os.windows: implement <timeapi.h>

[0x08088405]

This PR adds bindings to the minimal timeapi.h interface, which is very useful for configuring the system timer resolution to get accurate Sleep() and such. Since this API is part of WinMM, a new file has been created for it, following the current pattern of one file per system library with the two basic WinMM types that timeapi.h depends on added as well.

I've also tested all the functions just in case since there's very few of them.

[lithdew]

For the tests you've done, is it possible to include them since the API surface is small?

[0x08088405]

For the tests you've done, is it possible to include them since the API surface is small?

Do you mean in the form of a Zig testcase? The actual results are entirely hardware dependent, if that's what you're asking.

[lithdew]

For the tests you've done, is it possible to include them since the API surface is small?

Do you mean in the form of a Zig testcase? The actual results are entirely hardware dependent, if that's what you're asking.

How about a smoke test just to check that the methods themselves can be used?

[0x08088405]

For the tests you've done, is it possible to include them since the API surface is small?

Do you mean in the form of a Zig testcase? The actual results are entirely hardware dependent, if that's what you're asking.

How about a smoke test just to check that the methods themselves can be used?

I looked in the other std.os.windows.* files and they didn't contain any "smoke tests" either, so I'll withhold from that for now, but can do if a collaborator requests such a thing. The issue is calling them at all adjusts your system timer and this can actually be system-global before Win10 2004, I don't think a test should mess with that.

[Vexu]

Looks like this should be exposed in lib/std/os/windows.zig.

[0x08088405]

@Vexu You're totally right, my bad. Actually committed referencing the file now.