@hughbe any progress on this one?

Yup I've submitted a bunch of PRs. This is a bit of an umbrella issue unfortunately - the more tests I add the more differences I find - so not sure how well it fits into milestones etc.

Got it.
When do you expect to reach the point of "good enough" coverage? Do you have rough estimation?

I've gone through HttpListenerRequestStream, HttpListenerResponseStream, HttpListenerRequest and HttpListener.
I'm probably going to limit myself to HttpListenerResponse and authentication to conclude.
Maybe a good idea for me would be to close this issue once all the active issues are fixed - I dislike umbrella issues as much as anyone

Sounds great, thanks!

@hughbe What other issues is this one tracking?

The main remaining issue this tracks are (assuming dotnet/corefx#19138 and dotnet/corefx#19219 are merged):

• GetClientCertificateAsync_NoCertificate_ReturnsNull: this test hangs forever on Unix. It's odd, as a call to BeginGetClientCertificate followed by EndGetClientCertificate works absolutely fine.
• Add_InvalidPrefix_ThrowsHttpListenerException_Windows: this is an argument validation thing: we can add http://192./ as a prefix.
• EndGetContext_InvalidAsyncResult_ThrowsArgumentException: no argument validation is performed on asyncResult. Needs a bit of refactoring to fix, but is a good chance to consolidate windows/managed.
• Write_TooMuch_ThrowsProtocolViolationException: no exception is thrown in managed. This looks tricky without deeper investigation, as it seems the Windows implementation tracks the content length specified in the response. Again, could be an opportunity to consolidate code here.
• Write_TooLittleAsynchronouslyAndClose_ThrowsInvalidOperationException/Write_TooLittleSynchronouslyAndClose_ThrowsInvalidOperationException/: no exception in managed. This is similar to the one above. On Windows, we throw an exception trying to dispose the response stream if the response stream is incomplete.
• EndWrite_InvalidAsyncResult_ThrowsArgumentException/EndWrite_CalledTwice_ThrowsInvalidOperationException: managed implementation performs insufficient argument validation on the asyncResult. This is a chance to consolidate the two implementations more.
• EndRead_InvalidAsyncResult_ThrowsArgumentException/EndRead_CalledTwice_ThrowsInvalidOperationException: same as above
• Read_FromClosedConnectionAsynchronously_ThrowsHttpListenerException/Read_FromClosedConnectionSynchronously_ThrowsHttpListenerException: no exception thrown. I don't know why.

This is almost certainly too much work for before ZBB

Networking has already exception from ZBB.
@Priya91 can you please help prioritize the failures above? Which are likely impactful enough for us to take them?

@Priya91 ping?

@karelz Looking at the tests, will update when I have more info. I would prioritize the issues with different platform behaviors, that is cases where exceptions are not thrown/thrown.

I don't think we need the umbrela issue anymore, let's consider it addressed.

@karelz, there are 14 tests disabled against this issue, and PR dotnet/corefx#19880 disables another 36 tests against it.

Ooops, thanks for catching that. I have moved the issue to Future. I assume we already made our peace with the disabled tests, right?
Or is there any 2.0 work left here?

I assume we already made our peace with the disabled tests, right?

I wouldn't assume that. Most of these differences we're only just learning about as @hughbe writes the tests we never had before. To quote you from 5 days ago:
https://github.com/dotnet/corefx/issues/19744#issuecomment-301279350

Are you saying we didn't have time to look at the failures yet? That would be concerning ...

They're marked against this issue, and you assigned this issue to hughbe. He's writing a ton of good tests and finding issues. To my knowledge no one has further investigated those issues, but you would know better than I.

I think I mixed it up with "we need tests" - @Priya91 can you please review the test failures and help prioritize them?
We need to identify those that are must-have for 2.0.

@karelz @stephentoub I've requested @hughbe to try to file separate issues for distinct known compat bugs, so that it's easier to surmarize what the issue is instead of reading the whole umbrella issue thread to figure out how much work is left. As to prioritizing the test failures, I'm not sure how to do that, all of them are differences in behavior between windows and unix, we need to fix all at equal priority for 2.0, since HttpListener was added for 2.0.

Closing this issue as I've separated all the issues out into their own tests in dotnet/corefx#19880
I'd like to grab some of the low hanging fruit this weekend, especially around HttpListenerResponse