Re: NEW ISSUE (?): LINK header

Mark Nottingham wrote:
> 
> See:
>   http://ietfreport.isoc.org/idref/draft-nottingham-http-link-header/

Yup. And: 
<http://greenbytes.de/tech/webdav/#draft-nottingham-http-link-header> :-)

> I still regularly get queries about the status of this; there are a fair 
> number of interested people, so I think it's a good idea.

Definitively; either stand-alone, or as part of RFC2616bis.

> The blocking issue is 'rel', as discussed earlier; in a nutshell, HTML 
> and Atom both have the concept of a link relation, but they have 
> different syntax and possibly different semantics.

The current state of things is confusing. We have:

- HTML4 defining a LINK element, corresponding to the RFC2068-defined 
header (or the other way around)

- no link relation registry

- the HTML5 draft (<http://www.w3.org/html/wg/html5/#linkTypes>) 
stepping in and claiming the registry (currently a Wiki (!) at 
<http://wiki.whatwg.org/wiki/RelExtensions>)

- Atom (RFC4287) defining a registry 
<http://greenbytes.de/tech/webdav/rfc4287.html#rfc.section.7.1> for Atom 
links elements.

> I've been thinking of writing up an I-D describing the problem, but as 
> of yet haven't had time.

Let's try start by agreeing on the problem then.

IMHO...:

- The HTTP Link header allows to express relations between resources. 
The type of the relation is specified by the "rel" and "rev" parameters.

- The type of link relations in general is orthogonal to the type of 
document it occurs in. So it's not limited to (X)HTML and Atom feeds. 
Thus, a generic registry seems to make most sense.

- For experimentation, it might be a good idea to be able to mint 
relation identifiers without registering them centrally.

Finally, I'm not convinced that we have a syntax issue between HTML4 and 
Atom; I think we discussed this some time ago over on the mailing list 
(can you remind us what you think the problem is?). Right now, 
<http://www.iana.org/assignments/link-relations.html> doesn't seem to 
list anything problematic, and if there is a syntax conflict, we 
probably can have a small document updating that part of Atom.

Best regards, Julian