- From: Mark Nottingham <mnot@mnot.net>
- Date: Fri, 16 Feb 2007 21:57:50 +1100
- To: Ian Hickson <ian@hixie.ch>
- Cc: "Roy T. Fielding" <fielding@gbiv.com>, "ietf-http-wg@w3.org Group" <ietf-http-wg@w3.org>
There are at least two ways to read your preference; a) the Link header should be specific to HTML, and applicable to no other format b) the Link header's rel registry should be HTML's, and other formats have to follow that I don't know that either of these is workable, given that HTML and Atom both use link relations. Establishing a registry isn't inventing a 3rd namespace, as long as its initial contents are aligned with current uses. The biggest technical concern that I currently have is that having a combined registry leaves the door open for relations to be registered that won't be syntactically valid in HTML4, if the Atom rules are followed. If the HTML4 rules are followed, it would mean that the Atom folks would have to agree to an amendment of what a registerable link relation is -- but then again, they're going to have to agree to pointing to the combined registry anyway, so it's not that much more to ask, perhaps. On 2007/02/14, at 7:27 AM, Ian Hickson wrote: > On Mon, 12 Feb 2007, Mark Nottingham wrote: >> >> Oh - and there's also already an Atom link relations IANA >> registry, so >> it would require a certain amount of coordination. >> >> What do others think? Should there be a single, flat link relation >> registry? > > The previous spec just deferred to HTML4 for its definitions, do we > want > to not do that here? It seems like it would be dangerous to invent > a third > namespace (especially one that overlaps with existing ones), > especially > for a feature used as rarely as this one. > > -- > Ian Hickson U+1047E ) > \._.,--....,'``. fL > http://ln.hixie.ch/ U+263A /, _.. \ _ > \ ;`._ ,. > Things that are impossible just take longer. `._.-(,_..'-- > (,_..'`-.;.' -- Mark Nottingham http://www.mnot.net/
Received on Friday, 16 February 2007 10:58:05 UTC