- From: Jamie Lokier <jamie@shareable.org>
- Date: Tue, 29 Apr 2008 13:51:08 +0100
- To: Phil Archer <parcher@icra.org>
- Cc: Brian Smith <brian@briansmith.org>, 'Mark Nottingham' <mnot@mnot.net>, 'atom-syntax Syntax' <atom-syntax@imc.org>, 'HTTP Working Group' <ietf-http-wg@w3.org>
Phil Archer wrote:
> I am sympathetic to your proposal for foo-Links - it seems clear and
> your experiments add weight to it. However, it seems to lose out on
> flexibility. With Link, anyone can create a new relationship type by
> providing an absolute URI, it's only relative ones that are tied to the
> IANA namespace.
>
> Putting people in a tight corner will lead them to ignore some of the
> rules and I fear you'll soon see 'links' of the type
>
> dunno_about_registration-Links:
>
> so that clashes become rather more likely.
How about:
Generic-Link: URL=value
Which is required to be written in exactly that form, the URN denotes
the relation type, and the URN is always unquoted, absolute, and in
a canonical form.
It seems to satisfy the same easy parsing and substitution
requirements that motivate foo-Link, while providing the flexibility
of URNs for unregistered relation types.
-- Jamie
Received on Tuesday, 29 April 2008 12:51:53 UTC