On Tuesday, May 10, 2005, at 03:41 AM, Graham wrote:
Unless you can explain how multiple different feeds can be obtained via one URI
As I explained on a thread last week, rel=self doesn't necessarily correspond with the address the feed is being served from. Stop making this mistake.
Let's look at the proposed text from PaceFeedLink, the proposal that added "self":
atom:head elements SHOULD contain one atom:link element with a rel
attribute value of "self" and SHOULD contain a href attribute with an
absolute URI as its value. This URI SHOULD identify the feed and a
representation exactly equivalent to the feed SHOULD be available
through a HTTP GET on the URI. If the link has a "type" attribute, its
value MUST conform to the Atom MIME type as described in Section 10.
The atom:link element MUST otherwise conform to the definition of a
Link Construct in Section 3.5.
"This URI SHOULD identify the feed and a representation exactly equivalent to the feed SHOULD be available through a HTTP GET on the URI." I don't think the full meaning behind that sentence made it into the actual spec. Here's what's in the spec--in one place:
3. The value "self" signifies that the IRI in the value of the href
attribute identifies a resource equivalent to the containing
element.and elsewhere:
o atom:feed elements SHOULD contain one atom:link element with a rel
attribute value of "self". This URI identifies the feed and a
representation equivalent to the feed.Both the proposed text and the text that made it in say that the URI (or IRI) identifies or SHOULD identify the feed. The proposal says that the feed SHOULD be available from that xRI.
Was not "self" added largely/primarily to enable feed readers that didn't get the information about the IRI from which a feed was retrieved to subscribe to it? Did we not discuss standard behavior when obtaining a feed from a different IRI being automatically switching to the "self" value when accessing it in the future? See http://www.imc.org/atom-syntax/mail-archive/msg12176.html for example.
