xml-dev - Re: Random XTech observations

[Peter Murray-Rust <peter@ursus.demon.co.uk>]

At 13:07 03/03/00 -0500, David Megginson wrote:
>I'm sitting on AC 538 on the tarmac at SJC, waiting to push back and
>start the long flight home to Ottawa.  While I'm waiting, here are a
>few random observations from three days of XTech (I'm not going to
>mention specific papers, except to say that the overall quality was
>extraordinarily high):

I agree. [I am grateful to GCA for supporting me to go and give a
tutorial]. I think it probably warranted an even bigger attendance.
>
>1. 14-hour days are too long, even though the town halls were worth
>   it.

I was very impressed [though not surprised} at the willingness to listen to
ideas that were provocative or didn't fit with "my" perspective. The
session on new simple approaches [SML, RELAX] was imaginatively conceived.
It was a platform for those prepared to challenge the "complexity" of XML
(including Schemas) and for the challengers to be challenged. I think it
took courage for some of the presenters to take the platform and they
deserve our thanks.

Similarly several people gave live demos. As a proponent of this, my heart
went out to those whose machines or software failed at critical times!

>
>8. It's getting harder and harder to find an XML conference that's not 
>   mostly marketing presentations.  That's the inevitable price of
>   success, but it was fun to be at a conference like the ones from
>   two years ago, full of hard-core developers.  Thanks, everyone
>   (especially to the co-chairs, Tim and Jon).

Yes. The density of people I wanted to meet was very very high. If an
XML-DEV member has to save up pennies for an XML conference this should be
on their list. Whereas XML99 had a much higher proportion of business cases.


Pleasant company [Edd] prevented me from writing a similar thing on the
plane, but here are some additional thoughts:

Jon Bosak:
	Jon was rightly honoured with a plaque which (I think) read "XML father"
though I didn't see it. Jon's keynote was - as always - delightful to
listen to for the care, precision and economy with which he uses words. He
emphasised that the W3C was a research/development organisation, whose
staff were active experts in that area, funded by the members who wished to
see deliverables. As with other R&D orgs, the director has the final
decision. OASIS, by contrast, was an organisation for interoperability and
standardization and the staff were primarily administrative. He reminded us
that XML was only part of the process of e-commerce and that today's XML
languages - agreements - must be very carefully constructed as they have to
last for many years. How do we make and share these? Jon told us that most
of his time was now spent in this area and he had little formal connection
with W3C processes.

Sam Dooley - MathML/DOM
	Sam described the process whereby he and other IBM colleagues had built
Techexplorer for MathML based on the DOM. I think this was extremely
important and vital for anyone implementing a DOM. Since I have been doing
the same thing for CML we had a lot to share. My summary of our
conversations is:
	- writing a DOM for a technical DTD is not trivial; there is no "royal road"
	- Sam and co didn't get it right first time (which reassured me in some
ways...)
	- in the general case you may have to write a class for every element in
the DTD and an interface for every attribute

	- the more we can share our experiences, the better and more consistent
our designs will be.

Sam finished with an Export/Delegate mechanism between the MathML classes
and the DOM. Essentially (I think) he built a parallel tree to keep track
of parentage. Other people I spoke to confirmed this. Only for simple
textual applications, or where there is useful abstraction (e.g.
mathematical operator) is there likely to be a lot of re-use. I assume the
SVG implementers have also had to go through this?
	Lauren Wood (DOM WG) invited me to send in my thoughts on this issue to
the DOM WG, and I shall try to find time :-)
	*** any shared experience on DOM programming will be extremely useful. ***

XUL/weblets - see other posting

Frank Olken (LBL) presented about the need to support measurement units
(e.g. 12 Celsius). Lots of people agreed. It's a passion of mine and it
stops spacecraft crashing [well sometimes].

Town Hall Schemas. 2 hours on Schemas. Very useful. Not many tomatoes
thrown. [I asked them "please can we have a freeware schema-driven editor?"
This Twist-like impudence goaded Henry Thompson into suggesting that there
might be a daughter-of-XED [he didn't put it that way].  But we *really
have to have a schema-driven editor if we are going to test it out, don't we?]

Walter Perry enraged some, and delighted people like me, with the
suggestion that XML could be used for e-commerce without an agreed
vocabulary. The document is alternately created by XML additions from
vendor and purchaser until they agree to transact by more formal means. It
was a nice illustration of how - in the next few years - XML and humans
will be intimately mixed in the business process. 

Microsoft. Microsoft are publicly committed to making their XML tools
compliant to the specs and they have a timeline - including bug fixes -
which delighted many of us. Whatever our religious views about vendors I
publicly owe a lot to Microsoft's commitment to XML. Without the XML and
XSL functionality in MSIE we would find it very difficult to develop and
demo application (I had to use the commandline or JUMBO...). Everyone -
including MS - agree that the "glue" surrounding XML applications is still
pretty hairy, so we need concerted action by others as well.

XML Query Town Hall. This appeared to be taking us back to basics - queries
that are general enough to answer a wide range of questions. But some
considerable way down the track. I though "XQL" (based on early XPath) had
been deployed in quite a few products, and although it may have its
limitations I found it useful. Is there scope for a simple XML query
language NOW?

More generally there was a distinction between "documents" and "data". Jon
Bosak reminded us - correctly - that he ("XML Java and the future of the
Web") and others had always seen XML as supporting data. A modern
interpretation - which may be novel - is between the docucentric and
infocentric approach. Some folks propose that XML is really only about the
infoset and the XML1.0 spec is simply a serialisation of something more
fundamental. Others feel that XML1.0 expresses the nature of "documents",
including their assembly from components. The main point is that we must
not fight over this - XML works in many domains and this is nicely
expressed in - say - MathML's distinction between semantic and
presentational approaches.

On that front, I was personally very pleased to hear of several groups
using - or starting to use - CML. I also appreciated the need for
presentational chemistry and Henry Rzepa and I are addressing this. SVG has
come at just the right time - we now know that if we create something in
CML it will be straightforward to translate it to print (it wasn't before).
Publishing is a very large activity and we'd be happy to hear (off-list,
please!)  from people who need to address chemistry.

Once again, thanks to all.

	P.



***************************************************************************
This is xml-dev, the mailing list for XML developers.
To unsubscribe, mailto:majordomo@xml.org&BODY=unsubscribe%20xml-dev
List archives are available at http://xml.org/archives/xml-dev/
***************************************************************************