On 3 May 2005, at 12:49 am, Antone Roundy wrote:
Huh? I thought it contained a URI from which the feed could be accessed, and that if one were to get the feed from a different URI, the preferred action would be to fetch it from the "self" URI in the future.
No. The spec says simply this:
"The value "self" signifies that the IRI in the value of the href attribute identifies a resource equivalent to the containing element"
If an aggregator is reading the feed from the "wrong" URI, the web server can easily be set up to provide HTTP redirects on that address.
If an aggregator is subscribed to the "self" URI, then it knows what entries have been there, and can't be deceived by a feed claiming to have an entry from there. You're correct that there's no persistence, but I'd think most feeds are going to have fairly stable addresses. As for uniqueness, how are you going to get multiple feeds from the same URI? Different feeds may CLAIM the same "self" URI, but only one can actually come from there, so spoofs can be easily detected.
It's the claiming that's the problem. If the real URI must always match the rel="self", then every Atom feed must be discarded when the two don't match, and the request attempted again. That seems like hazardous behaviour and mixing of layers to me. It also means a feed document is only valid when accessed from its original location.
Graham
