-----Original Message-----
From: AndrewWatt2000@aol.com
[mailto:AndrewWatt2000@aol.com]
Sent: Friday, May 09, 2003 1:58 AM
To: Michael Rys
Cc: elharo@metalab.unc.edu;
xml-dev@lists.xml.org
Subject: Re: [xml-dev] Ten new
XQuery, XSLT 2.0 and XPath 2.0 Working
In a message dated 09/05/2003 08:57:54 GMT Daylight Time,
mrys@microsoft.com writes:
I would also like to
point out that the WG takes critique seriously if it comes in through the
proper channels,
Michael,
In my experience over several months that is fair comment, although
procedurally I guess the WG has to take comments seriously. In general,
feedback from the WG has been patient, helpful or factual even when I sometimes
asked less than perceptive questions.
although at this stage, there will be a big reluctance to change the language
unless
there are real bugs (such
as the missing static type to denote untyped complex types) or it can easily be
done.
I did sense a tinge of reluctance in Michael Kay's rowing boat analogy. :)
However,
also note that there are several people that see XML Schema mentioned
in the documents and just
shut down.
<grin/>
I guess that is understandable. It is fairly intimidating to come to ten XQuery
/ XPath 2.0 / XSLT 2.0 documents (~600 pages??) and then to have to go
exploring W3C XML Schema specifications - that well-known model of
clarity - as a normative reference as a bonus!
I can see that from an implementor's point of view the question might be
simple, "Can I create an XSLT 2.0 processor that doesn't use W3C XML
Schema and PSVI?". The answer is equally simple, I think,
"No".
If you would look at the documents closely, you would find – with some
bugs as
the one mentioned above
– actually works very well on just untyped data. The only major
difference in XQuery 1.0 then to the XPath 1.0 behaviour is that one needs to
explicitly extract a singleton for function arguments...
I have made the point already on public-qt-comments but I think it bears
repeating. Currently it is very difficult for a "doubter" regarding
the wisdom of the W3C XML Schema embedding to find a *clear, concise*
description in any of the documents of how the author of an XSLT stylesheet who
doesn't want to use typing will find their stylesheet behaving.
I appreciate that it may be clear in the WG's eyes. But there are other
perspectives, viewed from which it is significantly less clear. I have spent
many hours in these drafts over the months and I have some sense of where to go
hunting to cross check pieces of the jigsaw. However, I don't think it is
realistic to expect an XSLT stylesheet author coming new to these documents to
be able to glean the key facts for them about how their untyped stylesheets
will and won't run. Something that is clearer, more concise and to a degree
pre-digested needs to be provided, in my view.
It seems to me that it is in the WG's own interests effectively to communicate
to a wider audience what they *have* achieved, given the fundamental decision
re W3C XML Schema.
My sense at the moment is that the stylesheet author's concerns are, or will
shortly be, largely accomodated. The XSLT processor implementor's concerns seem
to fall into a different category. As I mentioned in an earlier post I see it
as entirely possible that a further cycle of simplification will take place, in
significant measure because of the imposition of W3C XML Schema.
Andrew Watt