Peter at ursus.demon.co.uk
Fri May 16 01:59:51 BST 1997
I've just come back from SGML97 in Barcelona and thought some personal
comments might be useful.
Firstly I was delighted to meet SGML geekdom in the flesh :-) It's marvellous
finding so many people who were hitherto virtual. I came away with a strong
feeling of community, and many thanks to everyone who made me welcome.
Without question, XML was the major theme at the meeting, though associated
areas such as DSSSL were also causing a lot of interest. The ERB has done
a fantastic job in getting this done so efficiently, quickly, and also
making sure that the world knew how important this was. It's now very clear
that the major WWW-related companies are taking a very active role in
exploring the potential of XML, and a recent posting to XML-WG has confirmed
Netscape's interest. [The XML-spec printing was sponsored by SUN and
I think most people who are not used to virtual working will underestimate
how much time the ERB has spent on the process. As Eliot Kimber said in
his closing address, they had to choose between having a family life or an
XML-life. There were literally hundreds of mails a week. NOTE: all mail
on the XML-WG gets read very thoroughly by the ERB - even if it doesn't get
formally answered at the time. It's only necessary to make a point once.
The ERB supplemented e-mail with weekly conference phone calls, and this
is how decisions were taken. Quite apart from XML itself, I personally
commend the efficiency of the ERB's virtual process and shall try to abstract
from it those aspects which make it successful. Clear initial guidelines
help, and a wider community which is well versed in abiding by a standard
drawn up under legal guidance :-)
Other points. XML clearly fills many different roles for different people.
It's clear that people who sell complex SGML applications see different
benefits (and some concerns) from those who see XML as the next step from
HTML. Taking too narrow a view might sometimes cause unnecessary conflicts.
Eliot described XML's position vis-a-vis SGML as low-cost/low-benefit versus
high-cost/high-benefit and stressed the need for the additional components
such as DSSSL, architectural forms, etc. (Personally I would put XML as
lowish-cost/medium-benefit :-) It's important not to argue HTML vs
XML or XML vs SGML as such arguments are often meaningless or based on
limited views. Both DTD-less and DTD-full applications will benefit from
XML. The *use* of XML falls in a spectrum with fuzzy borderlines.
It's clear that DSSSL has a great deal of impetus and the only question is
whether the ERB can work fast enough for everyone else's expectations.
There many other problems surfacing. How does XML interact with HTML?
(are there XML plug-ins, should XML DTDs contain subsets of HTML, etc.)
Strong typing, and APIs (both areas that XML-DEV could work on). And perhaps
most excitingly for some of us the concept of Information Objects which was
mentioned in several talks.
My understanding of information objects (which has been designed into CML)
is that documents will frequently contain 'chunks' from several different
sources. For example chemistry papers frequently contain maths; but there
is no formal syntax for combining two different DTDs in the same document
(watch this space...). IMO a robust DTD-less XML document will most
likely be an aggregation of well-defined information objects (i.e.
individually parsable against a DTD), but where every document would be
likely to have a differing formal DTD.
It's clear that XML is able to greatly widen the market for *ML. Since
*ML will increasingly be co-existing with Object technologies it's
important that the applications are well designed and interoperate cleanly.
One great benefit of *ML is that people who deal with documents can
understand the power of *ML, whilst they might well 'switch off' when
confronted with objects.
Implementation is also very rapid - many companies have - or shortly will
have - XML implementations. This will have high benefits - let's also
push for low prices :-). It's becoming even more important that this group
helps to create reference sites with test data, DTDs, etc. so that these
tools can be evaluated.
Once again, a lovely time.
Peter Murray-Rust, domestic net connection
Virtual School of Molecular Sciences
xml-dev: A list for W3C XML Developers
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/
To unsubscribe, send to majordomo at ic.ac.uk the following message;
List coordinator, Henry Rzepa (rzepa at ic.ac.uk)
More information about the Xml-dev