This is the original text which was eventually
edited, translated and published in the
April 1994 issue of
the German "Multiuser, Multitasking Magazine".
This article was a joint effort with Kersten Auel
of iX Magazine and Susan Liebeskind of the
Georgia Institute of Technology.
Only my portion of the article is presented here.
The X Technical Conference (XTC) is put on each year by the
X Technical Conference 94 (Frischer Wind)
Overall, this was one of the best X Technical Conferences
I have attended. Most of the talks complemented the papers
quite well. The vast majority of the talks had accompanying
papers in the proceedings this year (Bob Scheifler's talk on
the Consortium and R6 being the most notable exception). As
usual, the emphasis was on real research, future directions,
and work in progress.
This was also the first XTC in several years where no major
bashing of any vendor or the Consortium seems to have occured,
even in the BOFs. (This may have been attendance-related.)
This was also the smallest XTC in years. With only about
600 attendees (down from a peak of about twice that) only
those who really, really wanted to be there were present.
The main reasons for the small turnout, according to
informal surveys by the Consortium and myself seem to be
a tighter economy and more conferences from which to chose.
The fact that X is now mainstream, and far more information
is available, is also a likely culprit. (The inevitable
whining about holding a conference in Boston in January
was also heard, and properly laughed at.)
Highlights From the Talks
Extending Xt to Support CORBA-Based Embedding
Charles A. Price, SunSoft, Inc.
Chuck described an experiment at SunSoft involving both
making Xt widgets CORBA-aware and constructing Intrinsics-aware
CORBA objects. While the experiment did not go terribly deep,
the results were very helpful in identifying the problems for
anyone considering a real implementation. The presentation
struck an excellent balance between introductory material for
the uninitiated and the details of the experiment. As usual,
one of the prime culprits identified as a potential problem
spot was resource management. It appears that geometry management
and concurrency may be less problematic than might initially appear.
Xvan: A True Multiple Screen X Server
Peter C. Jones, Sun Microsystem Laboratories, Inc.
Peter described an attempt to provide a logical screen
consisting of multiple physical screens, in an effort to
bypass restrictions imposed by X in a multi-screen
environment (windows are normally restricted to one
screen). He described the approaches tried and the
problems encountered in each (primarily speed-related).
The final solution was general, produced neglible loss
in server performance, and is transparent to X clients.
Peter noted the server areas requiring rework, and
indicated that the code will be made generally available.
Experience with XIE: Server and Client, Past Present and Future
Ben Fahy, PhD, AGE Logic, Inc.
This brief overview of the X Image Extension (XIE) was
thorough and informative. Ben explained how the image
engine works, and gave examples of using the basic API.
Just how well this talk was done may be seen in the
fact that the primary question from the audience was
why the team left the "w" off the end of "Photoflo".
Design and Implementation of LBX: an Experiment Based Standard (Pushing an Elephant Through a Straw)
Keith Packard, Network Computing Devices
Keith did his usual excellent, entertaining job (assisted
by quips provided by Jim Fulton). Beginning with an overview
of extant enabling technologies such as PPP, the talk moved
quickly into the real requirements for LBX, with practicality
in the current network environment a high priority. Problem
areas were identified, and approaches verified experimentally.
The presentation included plenty of results showing what worked,
what didn't, and what remains to be determined. An implementation
of an LBX server was briefly described.
New X Font Technology for X11R6
Nathan Meyers, Hewlett-Packard Company
One of the most long-awaited extensions to the capabilities
of X has arrived. Since the release of X11R2, programmers
and users have been clamoring for more sophisticated font
support. Nathan's talk struck another excellent balance
between theory and practicality, with plenty of example
font capabilities demonstrated. The limitations and
potential problems inherent in a flexible font description
scheme were also well covered.
Kerberos Authentication of X Connections
Tom Yu, student, MIT
Tom was late, so Stephen Gildea began with a functional overview
of how kerberos was integrated into the X security mechanisms.
When Tom arrived, he described in detail how the new scheme was
implemented. This is provided primarily as an alternative to
SUN-DES-1, which requires not only DES access (as does
XDM-AUTHORIZATION-1), but Sun's secure rpc product.
This was primarily another review of what would change in Motif
with X11R6 and with Motif 2.0, other OSF/Motif news, and a time
for questions and answers. Nothing of earth-shaking import was
unveiled, and for once the audience seemed, if not thrilled,
then at least supportive rather than hostile.
The primary focuses of the X toolkit BOF chaired by Donna
Converse, who has had primary responsibility for Xt for some
time, seemed to be related to threads, failsafe X and requests
for specialized hooks of use to limited audiences. Additional
issues raised related to documentation (most people in attendance
wanted the the Consortium to provide both specifications and man
pages for the X toolkit, and to work closely with O'Reilly to
make sure the documentation is available as soon as possible
after a release. (There is a group of people who persists in
believing the O'Reilly documentation buggy, apparently related
to the fact that early O'Reily manuals included some of the
same bugs as the Consortium's early documentation.)
Donna also raised the question of how people would feel
about OSF taking over maintenance of the X toolkit. The
general consensus among the BOFfers was a resounding, "No!"
The primary motivations of this reaction were (1) fear that
OSF would (again!) modify (break) Xt for its own ends, and
(2) fear that Xt might one day no longer be freely available.
Daniel Dardailler of OSF responded that (1a) OSF's Motif is
the primary user of Xt (this is true in a commercial sense,
but ignores the freely available widget sets and applications
built on them), (1b) there is no guarantee that the X Consortium
won't break Xt (although history shows they work very hard not
to), and (2) X isn't really free, anyway - we all pay for it
through increased fees from vendors who support the Consortium.
Last updated: 10 July 1996
Copyright 1994 Miles O'Neal,
TX. All rights reserved.
[remove the "XYZZY." to make things work!]
c/o RNN / 1705 Oak Forest Dr / Round Rock, TX / 78681-1514