Windows.Win32.System.RemoteDesktop.IWTSVirtualChannelManager.CreateListener : pszChannelName should be a null terminated string instead of Byte*

[LeonarddeR]

Fixes #1185

Note, I'm slightly puzzled because the parameter is const char* in the source, therefore I'd expected this being emitted as PSTR already without this explicit intervention.

[mikebattista]

@sotteson1 / @chenss3 could you please review to understand if the tooling is behaving as expected here before we approve the manual override?

[sotteson1]

@sotteson1 / @chenss3 could you please review to understand if the tooling is behaving as expected here before we approve the manual override?

The problem is in the original definition, it explicitly uses const unsigned char * instead of a type we recognize as being a string (like PSTR or LPCSTR). So, it's expected that we would need to add more data to the emitter.

[LeonarddeR]

Should I close this pr and create a new one that patches the emitter to cover these cases?

[LeonarddeR]

I had a quick look at the tsvirtualchannels.h header in generation/WinSDK/RecompiledIdlHeaders/um and indeed, there the definition is _In_ const unsigned char *pszChannelName,. However, in tsvirtualchannels.idl, the definition is [in, string] const char *pszChannelName,. Also in tsvirtualchannels.h in windows SDK 10.0.22621.0, the unsignedness is missing.

[mikebattista]

Is there an infrastructure issue here or is "patching the emitter" just what you originally did by updating the emitter settings?

[LeonarddeR]

I tried updating the emitter by adding "const unsigned char *" to the several cases that emit a pstr, but that didn't changeanything, interestingly.

[riverar]

[...] So, it's expected that we would need to add more data to the emitter.

I believe @sotteson1's point was that it is expected to need to feed in a more precise type in as this PR is doing. I don't believe he was suggesting that anyone go try to modify the emitter.

That said, we could, as part of a separate task, do something with the __RPC__in_string annotation.

[BenJKuhn]

Sorry @sotteson1, I seem to have accidentally removed you from reviewers somehow when adding & removing myself to test something, and github won't let me add you back. Feel free to add yourself again.

[LeonarddeR]

I'm slightly puzzled, when I merge main into this branch, I don't get a merge conflict for scripts/ChangesSinceLastRelease.txt and somehow unrelated lines end up in the diff. So I reverted the merge of main for now.

[LeonarddeR]

Ah, it instantly became obvious to me that ChangesSinceLastRelease.txt was entirely rebuild since we had a release since my initial commit.