April 3, 1999 - Jack Harich
Two Days Together
This workshop was in Atlanta on February 26 and 27, 1999. We divided the time into two phases - Where we want to go, and how we're going to get there.
"Where we want to go" involved a detailed presentation and discussion of the UHR Technology Grid and the Bean Assembler. This was a pretty intense conceptual waving of the arms to explain and understand all this leading edge technology. Most of the group had done serious study of the BA or had actually used it, and so we got down to lots of nitty gritty. By the end of Friday, the first day, everyone was on the same wavelength and we had psychological group coalescence.
Saturday morning we had three short presentations to help us shift gears and put our "manager hats" on:
- Mike Panetta, Whiz Kid Extraordonaire of REALM, demonstrated Lego's Mindstorms. This is a robotics kit for ages 11 and up that involves an extremely well designed UI designed with help from MIT's media lab. It lets the user visually snap together parts representing logic to program a robot, but never looks like programming as we know it. Programs are then downloaded via infrared ports to a physical robot that you snap together from Lego like parts. There was huge applicability to what we were trying to do. Several people got excited about the potential. Christian vowed to introduce his young man in the family to Mindstorms.
- Paul Reavis, Design Mastermind of Partner Software, discussed and demonstrated their use of Declarative Knowledge and serialization like techniques in their product, using their own format. Of interest is each product sale involves extensive product customization, as much as several months of work. Thus configurability is the essence of what they offer to the market. He will be rewriting their product to use the BA as the core architecture to achieve rapid high quality configuarability, and playing a major role in the BA project.
- Britt Kinsler, CTO of REALM Technologies, addressed what they are doing with UHR and the BA. He shared how Parts Centric Architecture fits in with their plans. They have used the BA, have seen big productivity improvements, and are committed.
"How we're going to get there" involved developing a plan. Here are the raw meeting notes:
1. First we discussed Where We Want To Go. We kept referring to these notes as we delved into further documents.
2. Then we considered all that had to be done in the form of Action Items. This proved to be so many we dropped refinement of this document and switched to a higher level.
3. After a break we launched a futher mental assult on the massive task ahead of us by examining these items from a Cost Benefit Analysis viewpoint. This showed the amount of work in certain areas and its importance. We begin to organize into groups at the bottom of the page.
4. We ended up by doing what all serious Open Source movements do - We organized into Working Groups. These will carry on the job, including reorganizing themselves as necessary.
It's difficult to capture how we felt during those two days, but let's try:
This is a radical idea, an Open Source approach to assembling systems from parts. After all, the IDE vendors have been at it (code centrically) for years with Visual Basic, Access, Delphi, JBuilder, Visual Cafe, PowerBuilder, Visual Age, etc. How in the world can a bunch of developers do better? Well, that's what we talked about in a million ways. One reason we can do better is we're developers building tools for developers. We won't end up with a vendor centric monolithic monster that locks you in and forces you to digest and purchase upgrades whenever the vendor wants. Another reason is we're following an entirely different architecture with a slew of innovative tricks that, as far as we can tell, will turn how we all develop upside down. We're also going the Open Source route, which is not a magic solution but will help greatly in this case. Hmmmmm, strong stuff!!! Full speed ahead!!!
The workshop ended with people expressing how excited they were, how solid the concepts the BA rests upon looked, and how well the workshop went. We stayed at an amazingly high level most of the time, hardly every drifting into code or detailed technical discussions. This indicates this group is thinking at the higher levels needed to manage the future of UHR successfully. Jack Bolles returned to Chicago, Christian Cryder zoomed back to Seattle, Paul Reavis rode back to Athens, and the rest of us retired to Hot'lanta. In the next few days Michael Ivey completed setting up a dedicated server for us, complete with mailing lists for each working group. As we proceed, he's ready with CVS. There's lots to do....
Comments on the Workshop
Jack Bolles -
Christian Cryder - "I came to the workshop to evaluate the architectural integrity of the BA and its approach to UHR. I left with the conviction that not only is this approach fundamentally sound, but it's one of the most cohesive and complete efforts I've seen; there are still areas of immaturity, but I see no gaping flaws...I'm looking forward to working with it." - Also see Christian Cryder's Take on the BA.
Jack Harich - "We kept the workshop small deliberately, so only the best came and we were able to have a somewhat intimate family style discussion. Boy, did we have one! It went much better than I expected. This is a great bunch of people, sharp, with zero conflicts and no dominant egos. People are starting to understand and visualize how they can use all this technology. Individuals are starting to see where they can plug in to help. It was great to see folks proceed foward to the Working Groups approach."
Michael Ivey -
Britt Kinsler -
Greg Kreis - "A very promising sign, IMHO, was the willingness for a few senior developers to dig into the micro-kernel and others lower layer parts of the BA, to help repackage it and analyze dependencies. This is a process that will result in a very, very stable platform for all of the rest of the work."
Joshua Marinacci -
Mike Panetta -
Ash Raymond -
Paul Reavis -
Paul Tod Rieger -