- From: Poul-Henning Kamp <phk@phk.freebsd.dk>
- Date: Tue, 01 Jul 2014 09:02:12 +0000
- To: Mark Nottingham <mnot@mnot.net>
- cc: "William Chan (ιζΊζ)" <willchan@chromium.org>, HTTP Working Group <ietf-http-wg@w3.org>
In message <6B92EBCF-4D52-4092-A4E8-56BFD5D1B25F@mnot.net>, Mark Nottingham writes: >Keep in mind that you're arguing on the side that's saying they won't >correctly implement an open standard if this feature is included 1. There's no standard, there's a draft. 2. I've never promised to implement HTTP/2 at all. I've said I'll do it if it proves worthwhile for me and my users. 3. If and when I implement HTTP/2, I'll have to balance the concerns of my users against being 100% compliant with bad ideas in the standard. For instance the current draft mandates DoS-vulnerabilities like this: "Any number of CONTINUATION frames can be sent on an existing stream" Varnish is never going to "correctly implement" that, *ever*. >BCP83 goes on to say that "good faith disagreement is a healthy part of >the consensus-driven process." You're welcome to disagree, but please do >so in good faith. I participate in good faith: Even before the first draft were launched I pointed out the most severe of the shortcomings it still suffers from, and from the very start I challenged the accelerated time-schedule proposed which left a messed up prototype as the only realistic candidate. What seems obvious, is that other concerns, whatever they might be and whoever might harbour them, are far more important than producing a good protocol that will do its job well for a long time. Yes, I still think that the HTTP2 draft should get Last Rites rather than Last Call. But failing that, I want see major bogosity, such as CONTINUATION and all the priority crap moved to extensions, so they can be negotiated, avoided and get a chance to prove their own worth in open competition with other proposals. Even more I'd like to see a simple sane change which would simplify and shorten the protocol specification dramatically, increase the performance significantly and overall increase the chances that implementing HTTP/2 would be worth my or anybody elses time. Poul-Henning -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 phk@FreeBSD.ORG | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence.
Received on Tuesday, 1 July 2014 09:02:37 UTC