- From: Mark Nottingham <mnot@mnot.net>
- Date: Wed, 18 Feb 2015 12:48:00 +1100
- To: Bjoern Hoehrmann <derhoermi@gmx.net>
- Cc: "Julian F. Reschke" <julian.reschke@gmx.de>, HTTP Working Group <ietf-http-wg@w3.org>, Barry Leiba <barryleiba@computer.org>
Bjoern, > On 18 Feb 2015, at 11:47 am, Bjoern Hoehrmann <derhoermi@gmx.net> wrote: > > * Mark Nottingham wrote: >>> On 18 Feb 2015, at 2:38 am, Julian Reschke <julian.reschke@gmx.de> wrote: >>> >>> I still believe we should actually *fix* the table. >> >> As we discussed (but for the benefit of everyone else) -- we've already >> talked about this extensively, and came to consensus that making such a >> change would be more likely to cause problems, without making a material >> improvement in the protocol. > > Do you have a reference for this conclusion? I do not really agree that > we should ship a new protocol with known bugs for the short-term benefit > of experimental implementations, and I would like to find out what I am > missing. I also note that the text Julian quoted needs to be changed re- > gardless of whether the table is "fixed". The relevant issue is: https://github.com/http2/http2-spec/issues/587 Note that was raised as a "non-blocking" issue -- i.e., even when it was raised, we acknowledged that it was only worth doing if other things were changing too. Here was the resolution: http://www.w3.org/mid/E465C1C7-20DF-4F78-9936-9C914042920A@mnot.net ... and this is perhaps relevant context: http://www.w3.org/mid/B47FA4E6-6F91-44A1-8257-AE5086EF4DC1@mnot.net Of interest may be the experiments Greg ran: http://www.w3.org/mid/CAH_y2NEfOXWRtEbO+uUCKroW+NPGtyjqxNan3p5G+uFzuxxnCA@mail.gmail.com ... which show that there is negligible change in efficiency when the static table is reduced or changed. Cheers, -- Mark Nottingham https://www.mnot.net/
Received on Wednesday, 18 February 2015 01:48:33 UTC