the day started with another keynote speech titled the eclipse way. jeff mcaffer outlined the concepts behind the eclipse development process and the key features that make eclipse such a strong force in the opensource and business world. what stick to my mind is the word “continuous”. Eclipse uses a fairly agile development model that enables the project to deliver high quality in time. This is something that is very important, especially from a business point of view.

Even though i didn’t commit a paper for the rcp symposia, i was allowed to attend it anyway (thanks again). The symposia didn’t cover rcp development itself, but new requirements that should get implemented in rcp. It’s really interesting to see the different usage scenarios for which rcp is already used. Seems like rcp is the evolving unofficial standard for rich clients. This is probably owned to the fact that the .net world doesn’t offer something equivalent in terms of adoption and features. this is the impression i got from the discussion about .net integration into eclipse.
After lunch i continued my personal rcp track. Wayne gave a short presentation of the new rcp features eclipse offers in the new 3.3 development stream. More importantly i managed to get a “eclipse installed” sticker for my notebook. I love these geekish gimmicks (btw. why are free shirts almost always only available in l, xl and xxl?).
The closing plenary (each chair summarized the symposia results in a five minutes talk) included a few laughs and announced the free bear at in the foyer. I used the time to talk to mike and donald about the dali/jsr220 project merge and the problems that resulted from this. Donald was really interested about it even though he worked for oracle until the beginning of 2006. i will try to send him an email as a reminder and summary of that discussion. It’s all about lesson learned, and the outcome of this project merge surely contains a few of these lessons.
I was also able to discuss the possibility to implement oodbms support for dtp with john, the project lead of dtp. Basically it should be pretty straight forward. One would just need to implement the actual (meta) model of a oodbms in emf. This wouldn’t even require a lot of coding. Hopefully i can manage to get a commitment to work on this when i finished my bsc.
Speaking of it, i’m going to go over my outline with christian tomorrow to throw out a few topics. I really need to cleanup my outline, since it contains just too many topics.