Martin Duerst wrote:
At 17:16 05/01/25, Julian Reschke wrote:
>Also; I sypmathize with supporting IRI, but that shouldn't mean it needs to replace any usage of URIs
Every URI is an IRI by definition. So all URIs that are in use can be used with Atom without any problems even if the spec says we use IRIs. Replacement is definitely not the right term.
Let's disagree here. If the spec currently says "URI" and is supposed to say "IRI", we *are* replacing URI by IRI (even if every URI is an IRI).
>(for instance, I also think that the introduction into XMLNS 1.1 was a mistake).
Well, I might agree that it was a mistake. Observable behavior for most implementations (in particular XSLT implementations, where you most easily can test actual namespace behavior) allows IRIs in namespaces anyway, so this could just have been a clarification to the XMLNS 1.0 spec, rather than bumping up the number.
The big difference here is that XMLNS uses IRIs/URIs as identifiers and only for that. However, what is an XSLT that transforms Atom content to HTML supposed to do when it encounters a IRI which isn't a legal URI? For instance, it can't put it into an HTML href attribute without producing invalid HTML.
Best regards, Julian
-- <green/>bytes GmbH -- http://www.greenbytes.de -- tel:+492512807760
