RE: XRD <uri> vs. Link: anchor=

[Eran Hammer-Lahav <eran@hueniverse.com>]

This is why HTTP Link header supports 'anchor'. I am not sure if there is a reason why XRD should. I agree that it is good if links are consistent, but ATOM and HTML do not support 'anchor', or excluding it seems more consistent. We also didn't identify any use case for 'rev' in XRD which 'anchor' provides a replacement for.

EHL

> -----Original Message-----
> From: Jonathan Rees [mailto:jar@creativecommons.org]
> Sent: Thursday, December 10, 2009 12:22 PM
> To: Eran Hammer-Lahav
> Cc: xri-comment@lists.oasis-open.org; Mark Nottingham; Ben Adida
> Subject: Re: XRD <uri> vs. Link: anchor=
> 
> (Removing cc: mnot and Ben, who didn't ask to be included, so as not to
> spam them... maybe I'll touch base with them separately)
> 
> Sorry I got mixed up, but:
> 
> You have a URI U, do GET U, receive 302 or 307 Location: V Link: W.
> You may want to put information about U in resource W.  (U and V do not
> name the same resource, i.e. different properties may apply to the
> two.)
> 
> Jonathan
> 
> On Thu, Dec 10, 2009 at 10:56 AM, Eran Hammer-Lahav
> <eran@hueniverse.com> wrote:
> > But, if you have an XRD use case for 'anchor' please let us know.
> >
> > EHL
> >
> >> -----Original Message-----
> >> From: Jonathan Rees [mailto:jar@creativecommons.org]
> >> Sent: Thursday, December 10, 2009 5:11 AM
> >> To: xri-comment@lists.oasis-open.org; Eran Hammer-Lahav
> >> Cc: Mark Nottingham; Ben Adida
> >> Subject: XRD <uri> vs. Link: anchor=
> >>
> >> [ bcc: www-tag  for the TAG's information.  To broaden the discussion
> >> beyond XRD change the cc: list as appropriate. ]
> >>
> >> XRD <uri> seems to do exactly what the Link: header's anchor=
> >> parameter does [1]. It would be nice if the two used the same token
> >> to communicate this function.  As <uri> is not descriptive of the
> >> role played by the URI, I recommend updating XRD to match Link:, by
> >> renaming the <uri> element (or attribute, should you go that route) to be
> <anchor>.
> >>
> >> Jonathan
> >>
> >> [1]
> >> http://www.mnot.net/drafts/draft-nottingham-http-link-header-07.txt
> >>
> >> (p.s. I wonder if HTML5's <link> element should have an 'anchor'
> >> attribute, for symmetry ...)
> >