W3C home > Mailing lists > Public > public-grddl-wg@w3.org > February 2007

Re: GRDDL Link header [was: SW Coordination Group Meeting agenda, 2007-02-08]

From: Harry Halpin <hhalpin@ibiblio.org>
Date: Thu, 08 Feb 2007 21:41:19 -0500
Message-ID: <45CBDF4F.4040107@ibiblio.org>
To: Dan Connolly <connolly@w3.org>, public-grddl-wg <public-grddl-wg@w3.org>, SW Coordination Group <w3c-semweb-cg@w3.org>, Graham Klyne <GK@ninebynine.org>

Dan Connolly wrote: (cc'ing grddl-wg list)
> On Thu, 2007-02-08 at 14:00 -0500, Harry Halpin wrote:
> 1)  Use Ian's current suggestion to assume Mark Nottingham's draft will
> at some point be accepted as a RFC and keep the use of the both "Link"
> and "Profile" headers [2]
> 2) Change Ian's proposal to just use "Link" headers, which are already
> part of the approved IETF RFC 2068 [3] and so according to Graham can
> "legitimately appear in the wild" [1] The choice between (1) and (2) is design question. I don't expect
> to get any answers, let alone definitive answers, to design
> questions, in the CG.
It's not just a design question, since the WG concensus on closing the
issue is directly dependent on the possibility of getting this through
IETF action. Ian Davis's original e-mail [1] incorporates the *same*
information given in his proposal to the WG [2] using *only* a Link
header, instead of a Link and Profile header.

So it's a process question - what is the model for W3C interacting with
the IETF? If we are relying on a feature of HTTP headers as given by an
IETF RFC to solve an W3C WG issue, does the IETF RFC have to be the most
current RFC, or can it be an older RFC? If the answer is "older RFCs are
okay", then we can just use a "Link" header per Ian's original message
[1] and therefore not be dependent on IETF process. Otherwise, we have
to actively aid and abet getting Mark Nottingham's IETF Draft [3]
through IETF process in order to secure a new RFC with both Link and
Profile Headers, which I'm not sure if the GRDDL WG can spare the cycles
and/or the IETF process can reasonably be expected to be finished in a
less than a year for Mark's draft. Maybe or maybe not.

Graham seemed to suggest in his e-mail [4] that "older RFCs are okay" in
his suggestion:
"I think that what could be done now is to request a permanent registry
based on the RFC 2068 specification (based on that, Link headers can
legitimately appear in the wild, even though its not present in later
specifications), with a commentary noting that it has been dropped from
HTTP specifications and is being developed separately with a reference
to Mark's

In which case we do not *absolutely have to have* Mark's draft through
IETF process, but can go through the (I assume) shorter process of
getting a permanent registry entry for the Link (and not Profile) header
based on RFC 2068.

Of course, that would be a hypothetical course of action that would
require GRDDL WG concensus.


> I don't see any way to just use the Link header.
> The WG decided on (1). I haven't seen any new information
> that suggests re-opening the issue. Advice from the CG
> in the direction of 3) would be new information.
> Slight tweak: we shouldn't just assume mnot's draft
> will at some point be accepted. We should actively pursue its
> progress thru the IETF standards process. Danny A's msg
> is exactly in the right direction...
> http://www.w3.org/mid/1f2ed5cd0702080243l6a8b628aib67ae02c72e2eb4e@mail.gmail.com
>> [1]http://lists.w3.org/Archives/Public/public-grddl-wg/2007Feb/0052.html
>> [2] http://www.mnot.net/drafts/draft-nottingham-http-link-header-00.txt
>> [3] http://www.ietf.org/rfc/rfc2068.txt
>>        thanks!
>>              harry


Harry Halpin,  University of Edinburgh 
http://www.ibiblio.org/hhalpin 6B522426
Received on Friday, 9 February 2007 02:42:02 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:39:10 UTC