A. Pagaltzis wrote:

* Eric Scheid <[EMAIL PROTECTED]> [2005-08-05 06:00]:
ah, but how do you know those URIs are id's?

or more specifically, is the following URI an id or a location?

   <link rel="something-you-don't-know" href="..." />

You don’t know. Just goes to show once again that Antone was
right all along to rail against abusing atom:link for that
purpose.

Of course you could say that consumers should disregard links
with relationships they don’t understand – which, incidentally,
is equivalent to disambiguating whether atom:link/@href is just
any URI or supposed to be dereferencable. Other than those of us
discussing the threading extension, noone has opined about that
matter.

I'm all for taking the approach of having consumers disregard link relations they're not familiar with only I would word it differently. If a consumer comes across a link relation it is not familiar with, the consumer should treat the href URI as an opaque identifier and not assume that it is dereferenceable. Otherwise, the definition of the link relation should unambiguously indicate whether or not the href URI is or is not dereferenceable.

- James

Reply via email to