I'm not sure what you mean here. A particular namespace declaration will
have zero or one srcdef, no matter where on the web it is accessed from.
Rick's point is that it should allow zero or more.
> This is obviously required for
> firewalls, but could also be valuable if the implementation language in the
> src= file could vary (e.g. Perl, Java). Of course it shouldn't act as a
> pseudo-stylesheet mechanism.
I'm not sure that implementation language is the crucial factor. Schemas
will hardly ever be full computer programs, and I would be worried about a
system in which the exact same schema is expressed in two different
languages (especially two different programming languages!). I think that
Rick is referring to different levels and types of schemas...one for
validating the parse tree, one for advanced hyperlinks, one for
knowledge-representation assertions (e.g. RDF Schemas) etc.
The namespaces proposal requires us to choose a single schema as the
"canonical" one as if all of the other levels and types of schemas were
somehow less important.
Paul Prescod - http://itrc.uwaterloo.ca/~papresco
Three things are most perilous: Connectors that corrode
Unproven algorithms, and self-modifying code
http://www.geezjan.org/humor/computers/threes.html