Minor changes to fix some major issues on mobile when using gRPC.

[slightfoot]

• Fixes bug when a ResponseStream is cancelled the underlying HTTP/2 stream is not disconnected so the transport never goes inactive/idle.
• Add connectTimeout to socket connections.
• Fixes bug where connect state is mismanaged on socket error.

[thelinuxfoundation]

Thank you for your pull request. Before we can look at your contribution, we need to ensure all contributors are covered by a Contributor License Agreement.

After the following items are addressed, please respond with a new comment here, and the automated system will re-verify.

• User @slightfoot isn't covered by a CLA. They will need to complete the form at https://identity.linuxfoundation.org/projects/cncf

Regards,
CLA GitHub bot

[slightfoot]

Linux Foundation CLA signed.

[sigurdm]

This seems to solve a lot of problems!

We only need regression tests - are you up for writing some?

[sigurdm]

And thanks a bunch for contributing

[slightfoot]

Sorry, I just don't have the time.

[sigurdm]

No problem - I will look into this

[magillus]

Please merge or fix this issue. I was trying to solve issue when http2 connection restarted (due to local gRPC server restart) and auto connection.
I was going to prepare a PR then I saw this and applied that fix to my version - it works like a charm.

[slightfoot]

@sigurdm did you get around to the tests?

[sigurdm]

Sorry for the long delay.
I have tried to split this a bit - I created a pr with atest for the cancelling behavior.
I think we can should split this further into the connect timeout and the error handling. I think that would make it much easier to land this.

[VladimirLuzhin]

Hellow, is there any information when this pull request will be merge to the master branch?

[v0l]

This is badly needed..

[njovy]

When is it going to be fixed? This is fairly critical for streams.

[VladimirLuzhin]

Can you tell, when this pull request will be uploaded?

[slightfoot]

@sigurdm do you want me split out the error and timeout behaviour?

[sigurdm]

Yes, that would be lovely

[slightfoot]

@sigurdm removed the connection timeout handling. Its also up to date on master.

[slightfoot]

@sigurdm connection timeout is on separate branch.

slightfoot@11c9f42

[slightfoot]

@sigurdm poke.. shall we try to get this merged before it goes out of date again.

[slightfoot]

@sigurdm I just had someone reach out to me because they were caught out by the issues this ticket fixes. Can we get this merged?

[mit-mit]

@mraleph can you help with merging this if we get this brought up to date?

[mraleph]

I would be happy to land this @slightfoot! However this requires rebase and (regression) tests which validate fixed issues.

[mit-mit]

@slightfoot any chance we can get you to rebase and add tests?

[mraleph]

Closing as assumed stale for now.