Re: XML Core -> I18n Core: IRIs as namespace names?

Phillips, Addison scripsit:

> I note that XML Namespaces 1.1 2e actually does reference IRI (RFC
> 3987), while several of the XML Base specifications reference something
> much like an IRI (which has been given the name "LEIRI"). It's probably
> not a good idea to have both concepts running loose at the same time.

We aren't going to extend LEIRIs to anything that doesn't already
have them; hence the name "Legacy".  Indeed, the use of non-IRIs in
LEIRI contexts is discouraged: do you really want to see a DOCTYPE
declaration like

        <!DOCTYPE foo SYSTEM "<>">

(which is equivalent to the relative reference "%3C%3F")?  I sure don't.
But neither do I want to break backward compatibility by forbidding them.

> Also, I would point out that Section of IRI does address
> normalization, at least to some degree. Requiring NFC in namespace IRIs
> would address the composed-v-decomposed and combining-mark-reordering
> attacks (it would not address all visual spoof attacks---nothing
> can completely insulate a system from such an attack, however,
> even in ASCII). I know that does not require NFC (in fact,
> it requires that late normalization to NFC *not* be done during IRI
> comparison). However, namespaces could define non-normalized IRIs
> as illegal.

Excellent point.  I'll share this with the XML Core unless you object.

> That said, I think using IRIs for namespaces makes a lot of
> sense---especially if we allow elements, attributes, and so forth to
> use the full range of Unicode. And since XLink and other specs use
> (LE)IRI, it would make some sense to port it.


> That, of course, is my personal opinion with only a few minutes thought.


That you can cover for the plentiful            John Cowan
and often gaping errors, misconstruals,
and disinformation in your posts      
through sheer volume --that is another
misconception.  --Mike to Peter

Received on Thursday, 14 August 2008 00:21:01 UTC