- From: Brian Pane <brianp@brianp.net>
- Date: Mon, 15 Apr 2013 09:15:42 -0700
- To: "ietf-http-wg@w3.org" <ietf-http-wg@w3.org>
- Message-ID: <CAAbTgTsczdi3uN4fERxoyEc6zMQeMRQuvj00GqAkd=u_U1kVdg@mail.gmail.com>
I recommend keeping that bit reserved and leaving Stream ID at 31 bits for an additional reason: there are some very common programming languages with only signed 32-bit integers, and inattentive implementors who try to enforce assertions like "stream IDs must increase monotonically" will encounter surprises. Or, more likely, their users will encounter surprises, years later. On Saturday, April 13, 2013, Roberto Peon wrote: > This is for prioritization experimentation in the future. The bit allows > for priority level vs resource ordering without bloating the payload of a > reprint frame. > It was originally for control vs data. > > -=R > On Apr 12, 2013 11:50 PM, "Mark Nottingham" <mnot@mnot.net<javascript:_e({}, 'cvml', 'mnot@mnot.net');>> > wrote: > >> Looking at the minutes from Tokyo, this was originally for control vs. >> data (as in SPDY). >> >> I think there's been some discussion about discarding the control bit; >> OTOH, if people are going to define extension frames, it'd be nice for >> intermediaries to know whether they count against flow control without >> having to understand their semantics... >> >> Cheers, >> >> >> On 13/04/2013, at 3:43 PM, Adrien W. de Croy <adrien@qbik.com<javascript:_e({}, 'cvml', 'adrien@qbik.com');>> >> wrote: >> >> > >> > future proofing? E.g. if we need to move to another format or size for >> stream ID? >> > >> > >> > ------ Original Message ------ >> > From: "Brian Raymor (MS OPEN TECH)" <Brian.Raymor@microsoft.com<javascript:_e({}, 'cvml', 'Brian.Raymor@microsoft.com');> >> > >> > To: "'ietf-http-wg@w3.org <javascript:_e({}, 'cvml', >> 'ietf-http-wg@w3.org');>'" <ietf-http-wg@w3.org <javascript:_e({}, >> 'cvml', 'ietf-http-wg@w3.org');>> >> > Sent: 13/04/2013 12:48:46 p.m. >> > Subject: 3.3.1 Frame Header: Purpose of 1-bit reserved field? >> >> 3.3.1. Frame Header >> >> >> >> |R| Stream Identifier (31) | >> >> >> >> >> >> R: A reserved 1-bit field. The semantics of this bit are not defined. >> >> >> >> I was curious about the purpose for the 1-bit reserved field. Can it >> be deleted and the Stream Identifier increased to 32 bits? >> >> >> >> https://github.com/http2/http2-spec/issues/67 >> >> >> >> >> >> Brian Raymor >> >> Senior Program Manager >> >> Microsoft Open Technologies, Inc. >> >> A subsidiary of Microsoft Corporation >> >> >> >> >> >> >> > >> > >> >> -- >> Mark Nottingham http://www.mnot.net/ >> >> >> >> >>
Received on Monday, 15 April 2013 16:16:10 UTC