W3C home > Mailing lists > Public > ietf-http-wg@w3.org > January to March 2008

Re: Reviving HTTP Header Linking: Some code and use-cases

From: Harry Halpin <hhalpin@ibiblio.org>
Date: Wed, 12 Mar 2008 12:45:13 -0400
Message-ID: <47D80899.7090101@ibiblio.org>
To: "Roy T. Fielding" <fielding@gbiv.com>
Cc: Julian Reschke <julian.reschke@gmx.de>, ietf-http-wg@w3.org

Roy T. Fielding wrote:
> On Mar 11, 2008, at 7:05 AM, Harry Halpin wrote:
>> Julian Reschke wrote:
>>> No, I meant the link header...:
>>> Link: <http://example.com/grddl.xslt>;
>>>   rel="http://www.w3.org/2003/g/data-view/transform"
>> Yep, that would be another way to do it, but only *if* rel allows URIs.
>> I thought it did not. Am I wrong?
> It does not currently allow URIs.  The link relationship space is
> supposed
> to be small, universal, and standardized (just like HTTP methods). 
> That is
> why it is better not to allow an infinite number of relation types,
> but instead
> define a few types that allows an infinite number of relations to be
> understood.
> However, RDF pretty much screwed us all on that one, so the reasonable
> next
> step is to allow URIs and have all flat names be relative to the same
> link relationship registry as Atom.

As regards the notion of putting URIs in rel values, are the
intersection of SGML names and URIs is an empty sub-set? It seems SGML
names don't allow ":", so there's probably no overlap...so, is there a
reason *not* to allow URIs in rel?

Or should we invent another parameter, say "ler" that takes URIs and let
rel keep only SGML names?

Another option might be to say that rel always takes URIs, but if just a
sgml-name is given, it's just a short-hand for the "URI of Registred
Links" + "sgml-name"

    Just wondering...


> Note, however, that the following performs the same job as Profile
> without
> new protocol and without excessive bytes on the wire:
>   Link: <http://www.w3.org/2003/g/data-view>; rel="profile"
>   Link: <http://example.com/mymicroprofile>;  rel="profile"
>   Link: <http://example.com/grddl.xslt>; rel="grddl-rdf"
> Yes, there is some tiny possibility that two different profiles will
> use the same relationship name for entirely different purposes.
> I don't care.  If they can't be distinguished by context, then they
> shouldn't be used together.  Profiles are a crutch.
> BTW, it is my opinion that both Link and PATCH should be restored to
> the main HTTP specs.  HTTP is incomplete without them.
> ....Roy


Harry Halpin,  University of Edinburgh 
http://www.ibiblio.org/hhalpin 6B522426
Received on Wednesday, 12 March 2008 16:45:46 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 11:10:45 UTC