[stellation-res] OOPSLA news

["Mark C. Chu-Carroll" <mcc@xxxxxxxxxxxxxx>]

I meant to send this earlier, but better late then never.

Last week, I went to the OOPSLA conference, where Eclipse was hosting
a birds-of-a-feather (BOF) session and reception. At the BOF, I
presented a brief demo of Stellation.

I was *amazed* by the amount of positive attention and buzz there was
around Eclipse. I ran a workshop on monday, and nearly every attendee
was working on something related to Eclipse. The only presenter at
my workshop who talked about a tool integrated with an IDE that wasn't
exclusively Eclipse was the AspectJ team, which supports Eclipse,
NetBeans, JBuilder, and emacs JDEE. The monthly downloads of the Eclipse
version of their tools have consistently exceeded the sum of all the
other downloads. Even outside the workshop, when talking to people who
didn't know I was connected to Eclipse, Eclipse seemed to be one of the
main topics of conversation, and the buzz was almost all positive. (The
only complaint that I heard is that the latest release uses code-assist
popups very aggressively, and they're not as fast as they should be.)

The BOF session for Eclipse was absolutely *packed*. At the beginning of
the session, every chair in the room was taken, and there were two
layers of people lining the walls. (At a rough guess, there were about
250 seats available... There were at least 300 people at the BOF,
probably slightly more.) 

The session started with Erich Gamma giving a 1/2 hour demo that showed
off both the latest features of Eclipse, and how to use Eclipse to build
extension tools. He and his co-speaker (whose name I've unfortunately
forgotten) repeated many times that the goal of the Eclipse Tools
project is to make programming in Java as Smalltalk-like as possible. 
And based on what I saw, they're doing a great job. I've downloaded
2.0.2, and started using it, and I've already adopted a lot of what
they showed. (The new refactoring stuff is wonderful; quick-fix has
become a useful tool for everyday work.)

Next up was John Duimovich, demonstrating some work he's done on
building a Smalltalk IDE on Eclipse. It was a very early prototype,
but pretty cool. Interspersed with his demo were frequent complaints
that the main problem integrating Smalltalk with Eclipse was that 
Smalltalk really shouldn't be using source files.

After those two, the focus shifted to research projects. It was supposed
to be four 15 minute talks. First a professor from Carleton University
talking about using Eclipse for a Scheme environment for education;
then me and Stellation; then a professor from University of British
Columbia talking about a query tool for Java; and finally, the AspectJ
guys talking about AJDE.

The Carleton professor decided to talk about 3 other projects at his
University in addition to the scheme environment, and ended up taking
over 30 minutes for his supposed 15 minute talk. He was sufficiently
dull that about 1/3 of the audience gradually got up and left as he
droned on. Once he was finally done, the rest of us had
to rush, because the facility wanted us out of the room promptly. 

The Stellation talk went well, if extremely rushed. I got to do a short
version of a demo based on Jim's fine-grained prototype work. People
seemed excited and interested. For a 7 minute talk, it went over quite
well.

The JQuery talk was interesting. The guy behind it is a professor named
Kris DeVolder. He's got a very deep prolog-like language that has a huge
amount of expressive power. The current implementation is terribly slow,
but I ended up spending several hours talking to Kris, and he's very
interested in working with us on integrating some form of JQuery into
Stellation, both to let us use JQuery for generating VSFs, and for
helping him to improve JQuery performance by using the information
retrieval stuff that I've been implementing.

Last but not least came the AspectJ demo. I've seen this quite a few
times before, so I was kind of bored by it, but that's not a reflection
on the quality of the work of the demo - it's just that I've personally
seen it several times already. It's actually quite impressive. AspectJ
is a very cool language, and the AJDE *really* makes it shine. The
weakness of aspect-oriented tools is that they have a tendency to
increase the amount of scatter in your code; with AJDE, they have
extensive tool support to help you understand the AO system and how
everything is being composed; and they have it well integrated with
the debugger. 

Overall, the BOF was really good. I was very impressed by the work of
all the presenters, even the Carleton professor. (He was a boring
speaker, but the work was pretty good.) 

After the BOF was a reception. I stayed for about 2 hours, and had at
least a dozen different people approach me to talk to me about
Stellation. All of them were quite excited by it. 

The one consistent thing that people seemed to want was for Stellation
to consistently behave as an Eclipse plugin. What I mean by that is
people want the extension capabilities of Stellation to be uniformly
represented as Eclipse extension points, and to be able to use the
Eclipse PDE to develop Stellation extensions. What this would mean to
us is to turn even the core of Stellation into a true Eclipse plugin,
and Eclipse launcher for running the command-line tool. 

We've discussed doing this before. I'm now finally convinced that
we should really be doing it. The interest and excitement around
Eclipse is amazing, and Eclipse folks are very interested in 
Stellation. The best way to encourage that is for us move
wholeheartedly into the Eclipse plugin model. From what I
understand, it shouldn't be too difficult to do. And it won't
preclude the use of Stellation as a command-line tool, or the
use of command-line tools for developing the Stellation core.

	-Mark


-- 
Mark Craig Chu-Carroll,  IBM T.J. Watson Research Center  
*** The Stellation project: Advanced SCM for Collaboration
***		http://www.eclipse.org/stellation
*** Work Email: mcc@xxxxxxxxxxxxxx  ------- Personal Email: markcc@xxxxxxxxxxx