Re: Streams after receiving GOAWAY

On Sun, Nov 15, 2015 at 03:20:22PM -0800, Glen Knowles wrote:
> I'm well aware of what a normative MUST NOT means, I'm arguing that it's
> use
> in this case is to strong (or to general) - in no small part because
> violating
> it is undetectable by the peer.

It *is* detectable through the fact that it advertises a stream ID
that has possibly not yet been sent by the client and is not the
2^31-1 magic number either.

That said, I do think that there are valid use cases for what you're
trying to do and that it could have been planned from the start, had
anyone thought about it by then.

I really think you should write a proposal for an extension. This is
a perfect use case for extensions, it just requires the two peers to
agree with it and it does not affect the framing at all. Maybe your
extension will become the default mode over time, just like everyone
supports SACK, window scaling, and timetamps on top of TCP nowadays.

Willy

Received on Monday, 16 November 2015 06:34:01 UTC