test-win32metadata failing

[tannergooding]

The test-win32metadata leg is failing with no changes made on the ClangSharp side:

D:\a\1\s\artifacts\win32metadata\generation\emitter\obj\generated\WinRT.modified.cs(3398,17): error CS0101: The namespace 'Windows.Win32.WinRT' already contains a definition for 'RoErrorReportingFlags'
D:\a\1\s\artifacts\win32metadata\generation\emitter\obj\generated\Dhcp.modified.cs(1268,210): error CS0246: The type or namespace name '__AnonymousRecord_dhcpsapi_L1229_C5' could not be found (are you missing a using directive or an assembly reference?)
D:\a\1\s\artifacts\win32metadata\generation\emitter\obj\generated\Dhcp.modified.cs(1282,210): error CS0246: The type or namespace name '__AnonymousRecord_dhcpsapi_L1243_C5' could not be found (are you missing a using directive or an assembly reference?)
D:\a\1\s\artifacts\win32metadata\generation\emitter\obj\generated\Dhcp.modified.cs(1297,210): error CS0246: The type or namespace name '__AnonymousRecord_dhcpsapi_L1260_C5' could not be found (are you missing a using directive or an assembly reference?)

[tannergooding]

CC. @sotteson1, was there a change made to the win32metadata fork of ClangSharp that needs to be backported?

[sotteson1]

I thought master had everything I needed but I'll take a look. I might have missed something.

[sotteson1]

I found the fix for the problem and I submitted a PR for it. If I had tried to get win32metadata to use the nuget package I would have hit this right away! Next time it updates with this change I'll work on that.

[sotteson1]

Oh, although the first error is something different. In my rsp I had been excluding that enum like this:

--exclude
Windows::Foundation::Diagnostics::RoErrorReportingFlags
ABI::Windows::Foundation::Diagnostics::RoErrorReportingFlags

But to fix this new problem I had to do this:

--exclude
RoErrorReportingFlags

I'm not sure what's different that causes this with the latest ClangSharp changes. I must not have had them in my copy.

[tannergooding]

But to fix this new problem I had to do this:

That doesn't seem right. I'll take a look

[tannergooding]

@sotteson1, this will be resolved with #219.

The logic for generating the qualified name was changed to use . instead of ::. #219 updates things so that either works for remapping