Re: HTTP/2 response completed before its request

On 2 July 2014 12:34, Zhong Yu <zhong.j.yu@gmail.com> wrote:
> I feel that maybe the two directions of a stream should have
> independent RSTs, so that we can model more error cases.

I think that is unnecessary, and it might even be a big mistake.

An endpoint that doesn't want to continue sending can stop sending.
An endpoint that wants to stop receiving can cease sending window
updates (and buffer what it already provided credits for).  It needs
to do the latter anyway, since it can't do anything about what a peer
has already committed to send.

Would we develop rules around reciprocation for a half-RST?

Received on Wednesday, 2 July 2014 20:04:02 UTC