RE: ANN: SAX Filters for Namespace Processing

["David E. Cleary" <davec@progress.com>]

> XML Schemas didn't cause the problem, but they did introduce two
> features (local element types and unqualified child element types) which
> helped legitimize its practice. And believe me, a lot of people think
> that if a tool is available, it should be used.

Let's take this argument to an extreme. There are members of the XML
community who loudly argue that using attributes is a bad practice. XML
Schema could have taken this point of view and not created attribute
declarations. These people would be happy, but other wouldn't. But then, it
wasn't a goal of XML Schema to produce a subset of XML as some on this list
have been discussing for years. XML Schema had to be able to model as much
of well formed XML as possible. That task more correctly belongs to whoever
creates XML 2.0, and that probably will not happen any time soon.

Also, if alternative schema languages like Trex and Relax are unable to
model this XML, then not only do they have a much lower bar to jump over
than XML Schema, but I'd also be concerned about what else they don't let me
do in the name of best practices. And if they do allow modeling such as
this, then they are no more guilty than XML Schema, which IMHO means there
is no guilt at all.

> This argument is fine for users, who can choose not to use a feature.
> It's of no use to people (like me) who write schema-driven software. If
> it's in the spec, we have to support it.

If you are looking to subset XML, then you are correct that XML Schema is
not the tool to do it.

Dave