* Asbjørn Ulsberg <[EMAIL PROTECTED]> [2005-06-27 13:50]: > Rename the "atom:uri" element or change its type to a Link > Construct.
The problem with that proposal, even if it wasn’t too late to make any changes, is that, well, it replaces atom:uri with a Link Construct. A Link Construct has more semantics than the atom:uri element is supposed to. Firstly, the @rel attribute is unwarranted in this context, since there is only kind of relation the URI in atom:uri can have with the entity described by the Name Construct it’s in. Secondly, the @rel is necessary for Link Constructs because they are meant to appear in mulitplicity, while atom:uri is not. You could avoid these problems by removing the restriction to a maximum of one URI, of course. Then the Link Construct becomes an obvious choice. But that would require complicating the Name Construct spec and require significant additionaly complexity in Atom consumers – for what benefit? I don’t see any tangible gain in that. The basic idea to rename the element to something more descriptive is, of course, not bad. It might have been warranted to lobby for renaming atom:uri to something other than atom:link, or maybe for replacing it with a @link attribute on the Name Construct’s root element. But, as mentioned, the spec, modulo bugs, is a done deal at this point. And if the atom:uri element’s naming really turns out to be the biggest wart in the spec, then I’ll gladly suffer it. :-) Regards, -- Aristotle Pagaltzis // <http://plasmasturm.org/>