Backport fix on IAsyncEnumerable from 5.0 to 3.1 (PR 24926)

[tebeco]

fixes #25164

Description

Found from a stackoverflow question.
System.Linq.Async has a ToAsyncEnumerable method which returns a generic class that implements IAsyncEnumerable which didn't work with our reflection logic.
https://github.com/dotnet/reactive/blob/7ad606b3dcd4bb2c6ae9622f8a59db7f8f52aa85/Ix.NET/Source/System.Linq.Async/System/Linq/Operators/ToAsyncEnumerable.cs#L34

This is a backport of #24926

Regression?

no

Risk

same as #24926
I see no obvious Risk

More:

As @BrennanConroy explained:

I mean it's possible, it's just the servicing bar is unlikely to be met. https://github.com/dotnet/aspnetcore/blob/master/docs/Servicing.md#servicing-bar

This might be rejected, but at least the PR is ready as the bug was discussed about on Gitter
This is also on LTS (since 5.0 will NOT be LTS)
eg:
The team responsible for AuthN / AuthZ Handler usually refuse to work on non-LTS, so no net5.0 for us :(

Before we were assuming if the type was generic then it must be an IAsyncEnumerable<> or some non-streaming type which is what that check was doing. But that was a wrong assumption since you can have a generic type that implements IAsyncEnumerable<> instead of being an IAsyncEnumerable directly. We then fallback to the logic below that statement that checks all interface types for IAsyncEnumerable<>. If you take a look at the added test case then you can see the structure of a type that failed the check before.

[BrennanConroy]

I'm not convinced we should bring this to shiproom for backporting. There is a single customer asking for it and there is a workaround provided.

[halter73]

I agree this is fairly low value. But on the other hand, it's very low risk. I wouldn't change when we patch next because of this issue, but if we're going to patch anyway, I'd want to take this.

[BrennanConroy]

@tebeco Can you change the base branch to release/3.1 please.

[halter73]

This looks good but it needs to target release/3.0 release/3.1.

[tebeco]

Currently rebasing,
For my curiosity may I ask why 3.0 it it's now "out of support" and LTS are 2.1 / 3.1 and current it 3.1 ?

crap, my branch is based on master lmao, sorry, rebasing on release/3.1 (I suppose)
mehhhh

[halter73]

Sorry, I should have said release/3.1.

[tebeco]

sorry about that delay I did not check the base branch, In fact I'm surprised I managed to apply a diff on master while it was fixed in 5.0 :D
i feel stupid

[tebeco]

should be better like this
Sorry about that

[halter73]

In fact I'm surprised I managed to apply a diff on master while it was fixed in 5.0

I'm surprised too. I wonder how that happened.

[tebeco]

How bad can it go .... 😂😂😂

Look like I just selected the wrong base branch in GitHub but it was properly created
So in fact the surprise is that the CI was able to build that "diff"

[halter73]

It's crazy that GitHub generated that "diff" when this was targeting master. It was going to let us merge it too, but thankfully @BrennanConroy noticed it was targeting the wrong branch.

BTW, thanks for showing me that Git Extensions has a dark mode 😄 I really like it. Do you know if there's a way to make the title bar match? I know I could change it to globally show a dark "accent color" for every title bar, but that messes with other apps.

[tebeco]

It's crazy that GitHub generated that "diff" when this was targeting master

oO right ... something else went wrong and it's a bit hard to find out what was it ...

kudos to @BrennanConroy

Do you know if there's a way to make the title bar match

Nop, I knew something was off, but now I really see it WHYYYYYYYY

[tebeco]

(restoring branch / running it to see if test fails locally too)
can't tell from build log if related or not

can you confirm if these E2E runs on build.cmd or is there like build.cmd -e2e or so to be sure ?

[ghost]

Hello human! Please make sure you've included the Shiproom Template in a comment or (preferably) the PR description. Also, make sure this PR is not marked as a draft and is ready-to-merge.

[BrennanConroy]

The failures are due to some SameSite cookie issues due to Chrome updating. This PR wont pass until we merge some changes into 3.1 first. We'll re-run this when that happens.

[tebeco]

do I need to change the format of the template and remove all the "MORE" from the PR description for the "servicing model" ?

[tebeco]

unrelated
@halter73 you can check gitextensions PR 8435 (avoiding linking unrelated PR)

[wtgodbe]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[wtgodbe]

The 3 red jobs are from friday - weird that github is still showing them here. CI is actually green