Minutes of the HTTPbis Working Group meeting in Vancouver, December 2007 Submitted by Mark Nottingham * First session We started with a presentation from Larry Masinter, chair of the former HTTP Working Group, about the history of HTTP, with an emphasis on lessons learned in the previous process (see attachment). After a review of the charter, Mark Nottingham proposed a working process for the group (see attachment). Yves Lafon then summarised progress on draft-lafon-rfc2616bis. Roy Fielding then gave an overview of the partitioned drafts he submitted. After discussion, the sense of the room was to move forward with the partitioned drafts as a starting point. Afterwards, Julian Reschke gave a presentation (see attachment) of i36, the ABNF conversion. After discussion of different aspects, the editors were directed to proceed with a straight mechanical translation, with issues of style and readability that this may bring up to be dealt with separately. Issue 30 regarding LWS and headers was discussed; it was suggested to the editors that they use different BNF rules for parse vs. generate (e.g., LWS and BWS for "bad white space"), as long as the number of situations where this is necessary is small.  Other issues were discussed without significant progress. For scribed logs of the session, see . After discussion, the sense of the room was to use draft-sayre-http-security-variance as a base for the security properties document.  * Second session Alexey Melkinov kicked off discussion of the security properties document with a presentation (see attached). Further discussion centred around whether the group should attempt to revise RFC2617 (to fix interop issues / errata, not revise the mechanisms themselves) and the Cookie specification; both were deferred, although it was acknowledged that part of the HTTPbis work is revising the "framework" of authentication (e.g., WWW-Authenticate, 401). Cyrus Daboo gave a presentation to frame discussion of issue 22 (metadata on PUT and other responses). After discussion, it was agreed in the room that the way forward is to clarify intent that ETag is a response header, and add admonitions about its use to implementors.  Issue 81 (content negotiation for media types) was discussed; Larry Masinter agreed to write text that contextualises it and cautions against misuse. Discussion of issue 93 (repeating single-value headers) centred around whether HTTP should document error behaviours. The sense of the room was that the fact that they aren't specified because the variety of possible valid error handling behavours is too broad should be documented in the draft, which Roy Fielding agreed to do.  It was announced that Roy Fielding, Yves Lafon and Julian Reschke would share editorship of the partitioned documents, and that a -00 and ideally -01 draft of them would be available by the new year.  For scribed logs of the session, see .