* Brian Smith <[EMAIL PROTECTED]> [2007-09-16 23:00]:
> In particular, note that I wasn't suggesting to use <atom:link>
> for this, but rather a new element <entry-link>.
Ah.
But then you have a problem – how do you correlate links and
entry-links?
<link rel="related" href="http://example.org/2007/foo"; />
<entry-link rel="related" ref="tag:example.org,2007:foo" />
If these are the same piece of information, how can I know? Your
proposed `entry-link` would really need to allow for an `href`
attribute to handle this case.
Considering how little use is being made of the versatility of
atom:link, though, I doubt that any implementors will care about
esoterica like `via` links specified by Atom ID. In this respect
I think defining a new one-off element in a single-use-case spec
like RFC 4685 was the better choice, as it’s easier to relate to
people why they’d want to make the effort to support it.
* James M Snell <[EMAIL PROTECTED]> [2007-09-17 01:00]:
> It would likely be better just to write a spec that adds a ref
> attribute to the existing atom:like element. The new attribute
> would be considered foreign markup and ignored by existing
> clients. It would look something like:
>
> <link rel="related" href="..." ref="..." />
That won’t work.
There will be cases where you have an ID but no permalink and
therefore nothing to put in `href`, but `href` is REQUIRED so you
cannot omit it in valid Atom. And to switch that MUST to MAY
you’d have to bump the namespace.
Regards,
--
Aristotle Pagaltzis // <http://plasmasturm.org/>
