Carl Brown

> The reason we added the CB to Johannes original proposal was to add support for proper back pressure handling. Which is somewhat necessary for async based servers ... I...

So the idea is just that the application has the opportunity to pause until after `socket().send()`has completed? Ok, so let me think out loud here: * So the pathological case...

Status update: I did a naïve implementation of this at https://github.com/carlbrown/http/commit/5d503a3e9166956fdd4cd7774b8aa3fadb236432 and it didn't work - I got duplicated bytes in the output, so the completionBlock was getting called more...

> That should be a non-reason. I think very few people care what BlueSocket does. Just drop it and use something else. That's my plan - I just haven't had...

Referencing relevant comment from #81 : https://github.com/swift-server/http/pull/81#issuecomment-343612743

I updated BlueSockets to the latest 4.0 toolchain (see https://github.com/IBM-Swift/BlueSocket/tree/UnsupportedWIP/) to see if that was the problem, and the BlueSocket tests run clean on Linux under the latest snapshots, so...

It appears that the issue is happening because, in the test that passes, the attempt to close the socket fails (although the test passes). Here's the working output (with some...

Here is the output of two test runs on the same system, one from Swift3.1.1 and the other from last week's 4.0 Snapshot: [swift.test.311.txt](https://github.com/swift-server/http/files/1159937/swift.test.311.txt) [swift.test.40.txt](https://github.com/swift-server/http/files/1159938/swift.test.40.txt) In both cases, the code...

Possibilities: 1. It's possible the system calls are behaving differently, either: 1. The system calls do something different when being called from Swift4 than when called from Swift3. I'd rate...

As far as i know, it's never worked on Swift 4. It certainly hasn't worked on any of the 4.0-DEVELOPMENT snapshots that I've tried. It has worked on all the...