Re: Rules & Grammars

[Rick Jelliffe <ricko@allette.com.au>]

From: John Cowan <jcowan@reutershealth.com>

> XML Schema, OTOH: why *that* is so hard that I fear I'm unable.

I think XML Schemas specs are being written so that implementers can tick
off that they have implemented parts. Probably what is needed is for enough
simple schemas to come out to demonstrate the appropriate cliches. If the
Great Ox is correct, and understanding involves going from what is
well-known to what is less well-known, then probably many people will need
to see what the analogs of DTD constructs are in order to get the idea.
From the look of the pre-draft XHTML modularization spec, I think that will
be very useful for people.

When I am teaching beginnning XML Schemas, I don't even mention the syntax;
I start with the datatypes, go from there to type derivation ideas, and then
to the structural compents, and finally how schemas are located and how
validation is performed.  Audiences seem happy and positive.  I don't
believe XML Schemas is something that can be learned well from just looking
at instances--so first the ideas then the DTD equivalents and only after
that the details of syntax.

> If you pile rule-based stuff onto the already overreaching
> Tower of Babel, you might well get a pile of rubble.

Here here.   Another point is that the abstract objects being manipulated in
different paradigms have no connection with one another: there may be no
conceptual fit.  A pattern in Schematron is a grouping of nodes potentially
from all over the document;  a multi-key references is similar; complex
types are element based or attribute based.

Cheers
Rick Jelliffe