SGML: Observations on Seattle Conference
Subject: Observations on Seattle Conference
Date: Thu, 29 Aug 1996 10:04:07 -0500
From: Len Bullard <cbullard@HiWAAY.net>
Newsgroup: comp.text.sgml
---------------------------------------------------------------------------
[Observations on Seattle Conference, by Len Bullard]
Because there was a long thread attempting to
discover if the recent Technical Corrigendum and
alignment of the SGML standards could help
sew the Web back together, I am sharing some
of my observations from the recent GCA conference
in Seattle where some of the issues were discussed.
GoodNews: The alignment is real not cosmetic.
After a year of intense work, there is an adequate
basis for using DSSSL, HyTime and SGML
in application suites. To understand how this
works, one must understand the grove and
grove plan concepts. What we need now from
the working groups is the assurance that we
will not see big changes soon. Unless the
standard is very stable, no one should be
encouraged to build products for it.
The answer to whether this can be used
to sew the web together depends on the
efforts of the SGML On The Web working
group sponsored by the W3C. My intuition from
seeing the presentations at the conference is
that this is possible. The work on groves is
significant.
BadNews: It is apparent from the reactions
of those who attended the conference that
only a tiny fraction of our community understands
the technical corrigendum. Mail I have
received since the conference states the opinion
that the presenters may have done damage by
not being able to clearly and simply present the
TC results. This is unfortunate. The best
explanation I have for this is that the material
is too new and the number of participants in
the design work too small to provide anything
other than the "inside rant" on these subjects.
There is also the issue of working with unfunded
efforts, but that excuse does not play well to
a worldwide community used to free information
and free code. Unless a larger group understands
the technical implications of the concepts and
can discuss them, this party is over.
Good News: The DSSSL engine shown at the
conference worked stunningly well. For those able
to master the intricacies of DSSSL stylesheets,
we finally have a robust way to encode and share
rendering information among SGML applications.
Unless one has been an active SGML practicioner
for the last decade, one may not appreciate the
profound implications of this. More good news is
that this is well designed with regards to the use
of local libraries of DSSSL code that enable one
to use very small packets of information to transfer
the stylesheet initialization data.
Bad News: Just as the SGML community has
aligned itself, the W3C working group may drive
a wedge back into the body by trying to be the sole
arbiter of how these are applied to SGML On The
Web. I refer to the discussion of whether advanced
hyperlinking should be supported in DSSSL or
HyTime.
We must also be sure that any simplification
efforts do not result in a different metalanguage.
This is a stated goal of that effort, so I don't
have many concerns about it.
Good News: The SGML community can
lend assistance with the architecture to show which
roles each should play as envisaged by the designers
of the standards. Without this assistance, things
will be considerably slower and there will be a very
noisy fight. The real good news is that the cooperation
sought by many of us among the various groups chartered
to decide these things is happening. We must sustain this.
If we avoid company politics and the usual "more meta
than thou" problems, we can. Jon Bosak and Dan
Connoly are demonstrating strong leadership qualities
in this effort.
Neutral News: Efforts to get standards for scripting
for SGML/HyTime applications are moving again.
The MID demonstration showed that is was possible
to implement an SGML scripting application over a HyTime
engine using SGML syntax. The MID can be a good
sandbox for those interested in this topic. A MID
list sponsored by DISA has been created. The MID
documents are available as is the prototype created
for the US Navy. Source code is available on request.
IMO: A discussion of groves, grove plans and
how they unite the standards should begin on CTS.
This list remains the single thread that unites the
SGML community. While it is an unmoderated newsgroup,
it can still be the best medium for informing the
community of how this recent work can be applied
to our applications. If we do not use it, we have only
ourselves to fault. We must learn to focus our
threads around these issues and avoid thread drift.
This is difficult, but doable.
The time of doing all of our work in small groups cut off
from the main body of our members is over. It is said that
one cannot do hard technical work in large public groups.
This is wrong. It is just nerve wracking and noisy, but
ultimately, it is the way to assure that as a standard emerges,
there are running applications that use it.
If anyone has questions about the conference, I will
answer those that I can. Please ask these on CTS.
Len Bullard