Re: Comments on Extensible Prioritization Scheme for HTTP

Hi Glenn,

Thanks for reaching out.

I agree with Willy's words. I'll add that document is in AUTH48 now, and
technical changes so far in the process are best avoided unless there is
overwhelming evidence. Personally I'm not seeing that.

Your proposal focuses on frames but there's more to it than that when it
comes to priorities. It's complicated. HTTP/2 stream priorities are part of
HTTP/2, if you'd like that default off, it would have to happen there, and
that's asking everyone to change. It's too late for that now.

Furthermore, extensible priorities can work if the client uses headers and
no frames. HTTP doesn't require, or provide the means, for servers to
generically advertise the request headers they support. Additionally,
servers can ignore unknown frames, or parse the known PRIORITY_UPDATE frame
and ignore it's contents, per the spec.

It's noble to want a pattern but I don't think we have enough examples of
HTTP/2 extensions at the IETF to form consensus on one. So I think it's
more optimal to consider each extension on a case by case basis and apply
the judgements of the community.


Received on Thursday, 31 March 2022 12:07:10 UTC