<xml-dev@ic.ac.uk>
Notations don't require validating parsers, they only require parsers that
read and understand notation declarations. But they also require that there
be a DOCTYPE declaration so you can specify the notation declaration.
In this case, I think that a conventialized attribute would be sufficient,
but a notation of the same name as the value of the attribute should be
interpreted in the same way.
Or, said another way, an "xml:content" attribute could be presumed to be
declared with a value prescription of "NOTATION". The notation declaration
for the named notation is then implied.
This is analogous to "PI targets", where the spec says that the PI target
name is, semantically, the name of a notation, but you don't have to
declare the notation if you don't feel like it, but if there is a notation
declared with that name, it's rules govern the interpretation of the PI.
It would, I think, make sense to say the same thing for this.
And note that the notation of the data is, as previously mentioned,
orthoganal to how that data is encoded in the source document. Therefore,
you would not expect to have a notation of "base64".
Of course, if XML had data attributes, you could define a conventional data
attribute that specified the instance encoding, allowing different
notations to define what encodings their processors should support, but
nobody ever listens to me....
Cheers,
E.
-- <Address HyTime=bibloc> W. Eliot Kimber, Senior Consulting SGML Engineer ISOGEN International Corp. 2200 N. Lamar St., Suite 230, Dallas, TX 75202. 214.953.0004 www.isogen.com </Address>