John Panzer a écrit :
Ross Singer wrote:
... what argument are you making against using the http URI the
identifier? I understand if the tag URI is necessary for internal
reasons for the application, but if the http URI is, indeed, a viable
identifier for the resource... why not use that?
If the http URL is a permanent identifier that will never disappear
unless the resource itself does, and will never be reassigned to some
different resource, then sure. In practice it's very hard to
guarantee this.
Well it goes quite beyond the persistence of the value. The atom:id is
an opaque string, it's a IRI, not a URL. Bypassing the difference really
breaks the spirit behind it.
Use atom:link for that.
- Sylvain
