[
Lists Home |
Date Index |
Thread Index
]
> Is there a good argument why Namespaces-1.1 *shouldn't* be rolled
> into the XML-1.1 req/draft?
The message I get from the discussion is that no-one wants a new version of
the core XML specs unless it fixes some of the significant problems with the
current specs. I don't agree with Richard Tobin's policy (only do the things
that can be done quickly), because it means there isn't enough benefit in
the new version to justify the pain of having multiple versions around.
We already have configuration problems because of combinatorial effects (for
example, Saxon 6.5 doesn't work with Tomcat 3, because Tomcat 3 asks the
parser to switch off namespace processing and Saxon's version of AElfred
doesn't support that option). Introducing 1.1 versions of the specs, with
only very minor enhancements, can only increase that problem.
If we have a new version of core XML, then namespace conformance should
definitely be mandatory. Not because I like namespaces (on the contrary, I
opposed them vigorously when they were first proposed and thought the spec
so awful that it would never catch on). But nearly everyone is now using
them, and the minority who don't can always carry on using 1.0.
The problem is, what to do about DTDs. Here are three options:
1: kill them off (if version="1.1" there must be no DTD)
2: live with the current incompatibility between DTDs and namespaces
3: enhance DTDs to be namespace-aware
Personally, I don't want a new XML version that does (2). Either (1) or (3)
would be acceptable.
Michael Kay
Software AG
home: Michael.H.Kay@ntlworld.com
work: Michael.Kay@softwareag.com
|
