* Alan Gutierrez <[EMAIL PROTECTED]> [2005-12-18 06:10]: > I'm to understand that Atom has adopted the Tag URI as a a > perferred GUID for an Atom 1.0 feed. I've begun to use the Tag > URI in my work.
Tag URIs were advertised in Atom 0.3 days, but I don’t know that they’re currently suggested practice. Tim Bray has convincingly argued on more than one occasion that you should use HTTP URIs. RFC 4287[1] does not mention any scheme at all in normative language, and the informal examples in it use UUID URIs. (That’s what I use these days for my feeds.) > Eventually, I found myself parsing the Tag URI in Java, and > I've begun to wonder why the Tag URI scheme did not follow the > pattern of a hierarchical URI. It is instead an opaque URI. You’re not supposed to even want to parse IDs. You’re supposed to use identifiers as opaque tokens that do not convey any meaning whatsoever, other than to uniquely identify an entry. > Had Tag URI been specified as a heirarical, the parser would be > able to extract more payload from the URI. What for? All you use IDs for is to compare them to each other, and nothing else. [1] Damn, it feels good to be able to say that. Regards, -- Aristotle Pagaltzis // <http://plasmasturm.org/>
