xml-dev - Re: Improved writing -- who's going to pay for it?

[Rick JELLIFFE <ricko@geotempo.com>]

Linda van den Brink wrote:
> 
> Ronald Bourret wrote:
> > My point is that comprehensibility is critical, and that the
> > lack of it
> > is affecting things that are critical to them, such as reader input,
> > implementation experience, and rapid acceptance.
> 
> What I'm interested in knowing, is how sure are we that the w3c (schema and
> other?) specs are not comprehensible enough, and that implementation
> experience and rapid acceptance are being affected. Is it just a hunch we
> have? A general feeling among people on this list? What's the w3c's view on
> this?

I was disappointed that there was almost no comment on this in the Last
Call 
comments.  But please, in advance, can people be really careful not to
attack
either the W3C as some big evil giant forcing fat standards on you or
the
XML Schema WG (and editors) as people who don't care about careful and
systematic exposition to the target readership, to the best of their
abilities.


> If the XML-DEV community is convinced that Ronald Bourret has a point, then
> shouldn't we find out how right this conviction is and decide on possible
> further action?

Yes please.

I'll tell you why I don't think it matters so much.  Many stakeholders
are 
very keen that there should be no subsets of XML Schemas: strict
validity should
not be a function of the processor but the document.   Nevertheless, I
suspect that people who just want a DTD or XDR replacement (e.g. the
people
who would find RELAX core attractive) may find XSLD overkill.  So I
expect
that someone will create a version of XML schemas using a different
namespace and the same names, but simplified down to an XDR/RELAX Core
level: Key constraints removed, simple type derivation removed, facets
removed, include/ignore/redefine removed, complex type derivation
removed,
xsi:null removed, form/block/final/abstract removed.

A schema in this language could be converted into an XML Schema by
changing the namespace. Documentation and specs for it could be pretty
simple and easy to express, both informally and formally.  Of course,
we already have XDR, SOX, RELAX core etc, so perhaps we don't need 
another DTD-in-XML+simple-datatyping  schema language...but it would
certainly be possible and, I think, attractive.  Such a language
could even serve as a training wheels version for XML Schemas.

I don't expect W3C to do this. But there is nothing stopping some
firebrand on
this list coming up with it (the XSLT transformation to XML Schema
namespace would surely be trivial).  Then you can just ignore all the
bits of XML Schemas you don't like.  (Indeed, it is possible that
you could use redefine to even maintain the same namespace, but I
don't think that is fair use, and would only muddy the waters.)

Rick Jelliffe