* Sjoerd Visscher <[EMAIL PROTECTED]> [2005-08-21 13:40]: > Regarding the solution, my first suggestion would be to change > the xml:base to reference the atom document, e.g.: > > <link href="." xml:base="http://example.com/blog/feed.atom" /> > > This is also more consistent with the explanation.
Except it’s a @rel='self' link, so you really do want it to resolve to <http://example.com/blog/feed.atom>. @rel='self' is the one case where I still wonder how it should be handled. To be helpful and correct, you will want to provide an atom:feed/@xml:base for your feed, and likewise the conscientous will want to provide an atom:[EMAIL PROTECTED]'self'] for the feed – now regardless of *how* you write one or the other, if you use both as intended by their specs, then the atom:link is inevitably going to end up a same-document reference. I wonder if that’s really harmful, though. After all, @rel='self' is meant as the URL that should be used for *subscription* when the aggregator does not know where the document came from. This does not in itself imply dereferencing. (Semantics, semantics…) Further, I expect that in practice, clients will fall back to atom:feed/@xml:base when it exists but atom:[EMAIL PROTECTED]'self'] does not. In fact, I’ve been wondering whether atom:feed/@xml:base doesn’t obsolete the purpose of atom:[EMAIL PROTECTED]'self'], so that the former should have been the SHOULD that the latter is, and the latter not invented at all. It would seem that this notion would also be less controversial when backported to RSS2, as opposed to the item of including an atom:link in an RSS2 feed. Regards, -- Aristotle Pagaltzis // <http://plasmasturm.org/>