[
Lists Home |
Date Index |
Thread Index
]
>If we didn't have prefixes-in-content then we wouldn't have a problem.
This is hard to solve, given that
- we need to refer to namespaces in content (e.g. in XPaths)
- namespace names are URIs which are too long to type everywhere
- the shorthand we have for them cannot be recognised in content
below the application layer
If we were starting from scratch, we could specify a namespace or
qname shorthand syntax that was recognised and expanded by the parser.
For example, we could use curly brackets:
{http://example.org}foo
with a shorthand
{a}foo
recognised by the absence of a colon. Parsers could then expand {a}foo
to {http://example.org}foo (or to some kind of infoitem) and serializers
could convert back, though it would require a bit more to convert back
to useful prefixes.
>Putting aside wishful thinking about what might have been
Oops.
It seems that either the syntax must be interpreted below the
application layer, or the application must have access to the
information needed to interpret it, so I can't see any solution.
-- Richard
|
