* James M Snell <[EMAIL PROTECTED]> [2006-05-03 03:45]:
> Ok, so how's this (not word-for-word as I would write it in the
> spec, but you should get the idea)
>
> The ref attribute MUST be an absolute URI that matches a the
> resolved href URI of a replies link character-for-character
> (case sensitive)
>
> In other words;
>
> <link rel="replies" href="comments.xml"
> xml:base="http://example.org/foo"; />
> <thr:replies ref="http://example.org/comments.xml"; ... />
I don’t know if I like it. It really, *really* departs from
“something that’s as simple to implement as possible,” doesn’t
it? Not only is this just as hard for consumers to implement as
previous sketches, it’ll *also* annoy publishers, I think.
Question: what is the reason you are so staunchly opposed to
cloning `atom:link` for the extension’s purposes?
I’m starting to think that that is the only sane answer. If you
want to provide additional information about a link, you need to
associate this information with the link somehow. If you want to
stick to the spec mechanisms for extending the format, then you
have to provide this information in extension elements and you
must not add attributes to the link. So you must identify the
link by one of its attributes. The only attribute of `atom:link`
by which you can identify it at all is `href`. However, if you
factor in xml:base, you’re in a world of pain, and I don’t see
any way out of that.
The only other option I see is a very ugly hack: ride on the back
the `rel` attribute, abusing the fact that a relation may be a
URI to provide an identifier, ie. something like
<link
rel="http://purl.org/net/thread/replies?id=x3os882ja";
href="comments.atom"
...
/>
But that’s so awful and dirty on so many levels that that I have
to go wash my hands now. Ugh.
Regards,
--
Aristotle Pagaltzis // <http://plasmasturm.org/>
