W3C home > Mailing lists > Public > www-tag@w3.org > May 2009

Re: Comment on XSD 1.1

From: Rick Jelliffe <rjelliffe@allette.com.au>
Date: Thu, 14 May 2009 03:52:24 +1000
Message-ID: <4A0B08D8.1050203@allette.com.au>
To: Dave Peterson <davep@iit.edu>
CC: www-xml-schema-comments@w3.org, www-tag@w3.org
Dave Peterson wrote:
>  
> I see a similar reaction to XSD.  The folks who want something small 
> didn't
> participate in development, then complain about what comes out, and have
> once again built their own alternatives, which they seem to feel is what
> everyone needs (or should need?).  (N.B.:  That comment is not 
> particularly
> addressed at Rick Jelliffe; it's strictly a generic comment about a large
> group of people.)
Err, actually both Murata and I participated in XSD on the WG.  He went 
as far as resigning
because he felt it was going in the wrong direction. And I recall James 
Clark getting up
at the W3C annual meeting in Boston and clearly laying out his reasons 
why XSD was going
in the wrong direction.

James and I even made our own schema languages (TREX which fed into 
RELAX NG, and Schematron
respectively) a good eighteen months before the release of the XSD 1.0 
spec, to show that
other, simpler ways were feasible. (Murata's RELAX came out a fortnight 
after the  release of XSD 1.0)

And I have commented before that many of the long-standing complaints 
with XSD can be found
in the earliest of the public review, after the critical design 
decisions had already been made.

I spoke at WWW7 in Brisbane on the need for the mooted schema language 
to be based on
layers and plurality. A Microsoft representative spoke counter that 
idea, saying that
all we needed was a universal language. I think Jon Bosak chaired. The 
idea was not hidden.

The real issue is that the "technical people who wanted something small" 
thought they had made
their point with XML. They didn't realize it was being subverted by 
XSD.  Now there certainly
was a level of diverted attention from some: in the time when there 
should have been more public
review there was a large bunch of people calling for a simplified 
version of XML (SML, YAML,
etc): only people completely detatched from the world of markup should 
have been unaware
of the grassroots clamour for simplicity and rejection of complexity.

> It's hard to be all things to all people, and no matter what you do, 
> someone
> will step up and say you got it wrong.  C'est la vie.
But the WG is not being all things to all people. It is choosing to only 
deal with changes that
move XSD in the direction of being more complex-featured and refusing to 
deal with any
changes that profile or remove the features. The result may be a 
standard with less wrinkles, but
that is because it is a balloon.

> I wish the folk who
> think as Rick has described had chosen to be represented on the WG during
> the last many years; then XSD might well have been able to handle their
> desires as well.  Subsetting is not evil, it just needs careful 
> cooperation
> between the interest groups that need this or that subset of the whole.
They gave up when they saw (or felt) it was useless. I am sure there are 
better advocates
than James, Murata and I, and the hundreds of developers who have helped 
with
DSDL, but I gather they did not feel they could not make headway where 
we failed.
And I am sure that most and probably all members of the XSD WG have had 
grave
concerns about the complexity in XSD, probably none more than valiant 
Henry having
to formalize it all in words.

Of course, I want to fully acknowledge that the particular issues have 
been clarified
over the years. But honestly, if any person on the XSD WG in the early 
2000s was not
aware of these issues it was not because the information was not being 
aimed at their ears.

However, none of that matters. The point is not whether person X was 
right or wrong,
or person R was an incoherent ranter, or whether person Z didn't know 
the first thing
about markup language and was only there because their company had put 
them there
and would always vote with a block. The point is getting XSD back on the 
track it went
off a decade ago at its first draft. The issue is how to build in good 
software engineering
virtues of layering and modularity, and response to the initial broad 
use-cases, and how
to do it in a non-disruptive, positive, cheap, and productive way that 
can be win-win.

Cheers
Rick Jelliffe
Received on Wednesday, 13 May 2009 17:53:12 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:48:13 GMT