RE: <id> MUST be a URI (was Re: Editorial comments on mnot-03)

["Danny Ayers" <danny666@xxxxxxxxxxx>]

> 4.8 states that atom:id MUST be a URI.  An entry's <id> MAY be the same
> as the entry's <link>, but we're trying to discourage early adopter
> publishers from doing that so that early adopter developers don't get
> the wrong idea.  Not sure if or how to express that in a spec.

The current id/link setup seems to force a fork in the interpretation of
what a URI can be - one kind has a static representation, the other's
representation can change. I suspect this is probably a bad idea, as the
difference is only visible in the context of Atom, yet is saying something
considerably more general about the nature of those resources.

That it might be necessary to discourage early adopters from following what
would be the obvious practice should at least be considered a bad smell.

Cheers,
Danny.