[Marijn Suijten (MarijnS95)]

https://www.nuget.org/packages/Microsoft.Windows.SDK.Win32Metadata/60.0.34-preview

For me this most notably includes D3D12 Agility SDK 1.613.1 and annotations that leverage #2771.

[Kenny Kerr (kennykerr)]

Thanks for the reminder - I'll update the metadata shortly.

[Marijn Suijten (MarijnS95)]

Kenny Kerr (@kennykerr) why insta-close this when an external contributor has already put in the work? Don't you want non-Microsoft contributors in your commit history?

[Kenny Kerr (kennykerr)]

• There are many external contributors. The top contributors are all non-Microsoft last I checked.
• The metadata has blocking bugs that need to be resolved first (e.g. BCRYPT_HANDLE should not have a RAIIFree attribute win32metadata#1908)
• There isn't currently a way to validate the winmd provenance, so it has to be done in house. Once we have a yml workflow to validate provenance, external contributors can update metadata.
• Your PR snuck in unrelated changes beyond simply updating metadata.

[Marijn Suijten (MarijnS95)]

Can I submit the "unrelated changes" separately without risking them being closed, such as readability improvements to crates/libs/bindgen/default/readme.md and handling PSID?

---

For what it's worth, starting an insta-close with a more thorough review like the above would have been a lot more friendly.

[Kenny Kerr (kennykerr)]

• Discrete contributions fixing very specific things are welcome. A PR to improve a readme file for example is great.
• An "unrelated changes" PR that includes fixes to a variety of unrelated things will not be accepted.
• If there's anything that requires discussion, please open an issue for discussion to avoid spending time on a PR unnecessarily.

[Marijn Suijten (MarijnS95)]

• An "unrelated changes" PR that includes fixes to a variety of unrelated things will not be accepted.

Yeah, that's implicit. I shouldn't have sneaked the formatting fixes in for sure.

[Marijn Suijten (MarijnS95)]

and handling PSID?

This won't have an effect without the metadata update, and is probably not very useful to you in isolation.

[Marijn Suijten (MarijnS95)]

Kenny Kerr (@kennykerr) I do have a specific question with regards to DXGI_SWAP_CHAIN_DESC(1)::Flags (and the decode desc) which is of type UINT. In microsoft/win32metadata#1757 I associated that with DXGI_SWAP_CHAIN_FLAG, but enums have int ABI by default (unlike the enums I added in that PR in enums.json which were all given "type": "uint", and ILSpy shows e.g. enum DXGI_PRESENT : uint).

Is there anything we can do in windows-rs to tie this back to an enum type when the ABI mismatches? Is there a tracking issue open about this? Or should win32metadata override the enum ABI to be UINT (here and probably for others)?

[Rafael Rivera (riverar)]

[..] unlike the enums I added in that PR in enums.json which were all given "type": "uint", and ILSpy shows e.g. enum DXGI_PRESENT : uint).

We need to go back and fix that in metadata; looks like we missed these uint types. I probably should have been more clear in my PR comment.

[Kenny Kerr (kennykerr)]

We'll have to wait for this (and microsoft/win32metadata#1908) to get fixed before updating windows-rs.

[Marijn Suijten (MarijnS95)]

Rafael Rivera (@riverar) thanks, what's the right way to correct the type, list it in enums.json?

[Rafael Rivera (riverar)]

Marijn Suijten (@MarijnS95) Ideally yup, with the rsp method as a fallback. We can double-check with Mike too on the other side of the fence.

[Marijn Suijten (MarijnS95)]

[..] unlike the enums I added in that PR in enums.json which were all given "type": "uint", and ILSpy shows e.g. enum DXGI_PRESENT : uint).

We need to go back and fix that in metadata; looks like we missed these uint types. I probably should have been more clear in my PR comment.

Rafael Rivera (@riverar) just to double-check, what do you mean by this?

1. The type: uint was wrong, and should have been kept at the default int?
2. Forcing the untyped enums like DXGI_SWAP_CHAIN_FLAG mentioned here to uint, as that is how all current ABI is using it?

I assume the latter.