httpbis ietf 721 rfc2616bis draft overview ietf 72, dublin julian reschke mailing list: jabber:...
TRANSCRIPT
httpbis IETF 72 1
RFC2616bis Draft Overview
IETF 72, Dublin
Julian Reschke <[email protected]>
Mailing List: <mailto:[email protected]>
Jabber: [email protected]
httpbis IETF 72 2
How To Track Us• All issues are in Trac
– new issues discussed on mailing list– design issues opened by chair
• Drafts, including latest edits, are in Subversion– Checkins linked to issues
• Diffs on tools.ietf.org and http://svn.tools.ietf.org/svn/wg/httpbis/draft-ietf-httpbis/diffs/
• Each draft enumerates non-editorial changes in an appendix
httpbis IETF 72 3
Draft History (1/2)• 00 (December 2007)
– partitioned RFC2616
• 01 (January 2008)– changes from original errata list applied
• 02 (February 2008)– some work on BNF fixes, make each document have a
complete BNF– clarified requirements on PUT/201/Location– IANA media type registrations updated
httpbis IETF 72 4
Draft History (2/2)
• 03 (June 2008) – BNF: clarify HTTP-date (what to send, what to expect)– BNF: fix quoted-pair– IANA: header registrations, status code registry– Clarifications: Allow header, status 303 (“see other”),
PUT, charset quoting, Accept-Encoding qvalue default– Deprecate status 305 (“use proxy”)– Make weak ETags more useful for methods != GET
• „latest“ (July 2008)– Clarification on message length/connection closing,
OPTIONS request bodies– HTTP Method Registry
httpbis IETF 72 5
IANA: header registrations• Message Header registrations (according to RFC
3864) have been added (draft 03)
• Question: should we include additional information, such as:– general header vs request vs response vs entity– list syntax allowed– I18N (does RFC 2047 apply?)
– And if we do so, do we also want to modify the registration process?
httpbis IETF 72 6
Updating References• Mostly done
• Some downrefs for compression specs (RFC 1950..1952), see issue 68
• References to historic URI documents still in, expected to go away when we update to RFC3986, which in turn currently depends on BNF-to-ABNF conversion
httpbis IETF 72 7
IANA: Status Code Registry
• Previously defined in RFC 2817, now in Part 2 (as of draft 03)
• Basically keeps the registration requirements, but rephrases them according to RFC 5226
• <http://tools.ietf.org/html/draft-ietf-httpbis-p2-semantics-03#section-5.1>
httpbis IETF 72 8
IANA: Method Registry• New Registry
• In unpublished draft of Part2 – http://svn.tools.ietf.org/svn/wg/httpbis/draft-ietf-httpbis/latest/p2-s
emantics.html
• Registration requirements– identical to Status Codes– „IETF Review“ for new registrations– MUST state safeness– more required fields?
• Register methods not defined in HTTP/1.1 in a separate draft– work in progress– see
http://svn.tools.ietf.org/svn/wg/httpbis/draft-ietf-httpbis-method-registrations/latest/draft-ietf-httpbis-method-registrations.html
httpbis IETF 72 9
From 2616-BNF to ABNF• Not as simple as it seems• Need consensus how to deal with TEXT production, which
in theory allows RFC 2047 based encoding• Need consensus how to deal with Linear White Space and
the List Production– Clarify what RFC 2616 really says, vs. what is being implemented
• Status– Fixed simple errors (broken prose rules, name collisions)– Removed most prose productions– Started using RFC 5234 core rules– Expanded case-sensitive string constants to octet sequences– BNF, still in RFC2616 format, can be parsed by modified version
of Bill Fenner’s parser– Tools extract BNF and track the changes
httpbis IETF 72 10
I18N• TEXT production in theory allows RFC 2047
based encoding
• Observations:– I18N seems to be irrelevant for most of the headers in
RFC2616– Exception: Content-Disposition (which already has its
own escaping rules)– New specs work around the issue (for instance, Slug
header in AtomPub (RFC 5023))