Graham wrote:
On 31 Jan 2005, at 12:16 am, Robert Sayre wrote:
Graham wrote:
Yeah, that's a horrible loose end to leave hanging.
No, it isn't. URI comparison is not our problem, and what our spec says about it doesn't matter a bit.
Yes it is times one million. Ha ha I win et cetera
Defining rules for ID comparison is our problem if we're even going to pretend there'll be interop amongst Atom software. The spec currently says "Use these rules: [states rules]. Or use some other rules. We don't really care."
How about "Make sure your id is unique from a character-by-character perspective, but also unique in the face of scheme-specific comparisons". That is, don't lean on scheme-specific comparisons to match URIs, but they don't have to be canonical either.
a) http://Example.com/thing b) http://example.com/thing
a) is ok, but don't expect clients to match it to b).
Make sense?
Robert Sayre
