1. We are very much in line with the "hard problems" that
people are trying to solve in integrating the Web and Objects
to provide a technology for agents/infospheres.
2. In many cases, we've thought out the issues a lot more
than the people who attended the workshop. The workshop
itself was an identification of key issues, without much delving
into potential solutions.
3. We should get a subscription (and the CS library should get
a subscription) to the World Wide Web Journal. A press release
about the journal should be coming out on http://w3.org/ any
4. Some of the people I talked with:
A. Larry Smith of IBM has been doing workflow for years.
He says we should stay in touch with his group to leverage
off of his workflow models and scripting languages.
B. Sankar Virdhagriswaran of Crystaliz has been working
on a commercial version of agents for some time. He told me
first that the issues our group has been focussing on sound
very reasonable, and also that the application programs we
are considering (such as calendar, workflow, and shared document
systems) represent an excellent test suite for our ideas.
C. Mic Bowman from Transarc is a very interesting person.
We should look more carefully at his Synopsis File System, as
they've had to deal with many of the same issues we now are
D. Charlie Kindel and Nat Broman of Microsoft were very
clever guys. They were very attentive listeners.
E. Andreas Vogel gave an interesting tutorial on
Java ORBs. Basically, all Java ORB technology (there are at
least 10-12 products out claiming to have some kind of this
functionality) that currently exists seems very cumbersome to
use. People who write programs using Java ORBs could really
benefit from having some tools. We should install Blackwidow
on our system and play with it, though.
F. Some W3C people were telling me about Jigsaw and Jeeves.
We really need to carefully look at both of these products,
so we know what state-of-the-art in extending the server side
functionality on the Web is.
G. Someone (I forget who) told me to look at something called "Omni
Virtual Machine" -- but looking through Altavista, I haven't been
able to locate it. Oh well.
H. There's been some interesting distributed type checking
work done at Waterloo and NCSA, so we should keep tabs on those
projects as well.