Re: [xml-dev] Rethinking namespaces, attribute remapping

[Arjun Ray <aray@nyct.net>]

John Cowan <jcowan@reutershealth.com> wrote:
| Norman Walsh scripsit:

|> But if we allowed the PI target to be a QName...
| 
| It can't be.  But a PI target *can* be a notation name, which a NOTATION
| declaration then maps to a URI.
| 
| Ta-da!

Doesn't "can" leave room for false positives?  Wouldn't it be better to
have the PI target imply the *presence* of a notation reference within the
body, as per some internal *syntax* which it would be the purpose of the
target to identify?