James Holderness wrote: > Brian Smith wrote: > > Tatsuya Noyori wrote: > >> I like Atom syntax. And I think analogy is important. So I > hope that > >> the tombstones is similar in syntax to Atom syndication format. > > > > I agree. > > Do either of you have a technical reason for your preference? > Efficiency, interoperability, ease of implementation - > anything like that? Or is this just an aesthetic issue for you?
The main benefit is that the syntax and semantics can be completely delegated to RFC 4287 with minimal verbage. Even if you haven't read the tombstone spec, when you see an <atom:id> element, its syntax and semantics (especially the comparison rules and lack of a guarentee of dereferencability) are immediately clear. At the very least, I expect that the syntax of the timestamp to be consistent with RFC 4287 and RFC 5023, which means using "date construct" elements. It the timestamp is going to be an element then why not make the ID an element as well? RFC 5023 did a good job of reusing the RFC 4287 syntax whenever possible. I thnk that is a practice that should be followed by other extensions to RFC 4287. At the very least, such a policy will reduce the need to discuss syntax and allow more time to discuss semantics. http://www.informatik.uni-kiel.de/%7emh/curry/listarchive/0017.html - Brian
