Juraj Kirchheim

Threads are not a requirement though. The runloop could still make progress two downloads in parallel, so long as each of them is performed bit by bit.

Well, in theory the pipes drive the sources which each read a (possibly empty) slice through a worker which hands control back to the runloop, which then schedules the next...

This is not at all my strong suit. I was happy having something reasonably readable. If you feel that you can improve this, by all means, please do.

I propose to introduce some `abstract Ip(String) to String {}` for now and then we can add inspection utilities later down the road.

I am not particularly hurried for the inspection capabilities. My point was rather that we should introduce a specific type before 1.0.0 and then we can add any functionality later,...

Alright, thanks, I'll look into this.

Well, suspend puts a pending pipe into "standby" if you will. So that if you have many pipes the ones that have no data passing through, they don't compete with...

Ok, I have checked this, which was the most intelligible spec I could find: http://httpwg.org/specs/rfc7230.html#message.body My understanding is that if there's neither `Content-Length` nor `Transfer-Encoding`, there's simply no body. And...

On second thought if no such thing is specified for `POST`, `PUT` or `PATCH` we might still try reading for "a while" just so that non-conforming clients get through. Or...

If curl shuts down its end of the socket then that would cause an EOF on the server. There are some chances that the TcpClient will work for the same...