Friday, October 27, 2006

Another day at OOPSLA -tuesday

What to say about these keynotes? Hard to describe without having been there. Let's just say that they were as far away from code examples as you can get. One about very large scale systems and their structure, ability to change and adapt the other one about the pleasure in computing combining animism with computing.

This was followed by The Geography of Programming by Elisa Baniassad. Westerners think more of objects and easterners think more about relationships between objects. Quite interesting talk with some very good pictures and video to go a long with it. My own reflections. Obviously (or is it?) more suitable way of thinking for our modern systems, who it is getting increasingly hard to find all the dependencies between objects/modules.Relational languages instead of object oriented perhaps? Perhaps the move from relational databases object oriented ones should (also) be the other way around? Move the language to be relational instead of moving the database to be an object oriented one? How will it affect us, how should it effect us? In the near term maybe just think more about the relationships or roles between objects/components, or should we stop thinking/focusing about objects almost entirely?

A rather long lunch. Other than the usual suspects (Jon, Niclas, but no Jimmy) and some new found friends over here (Anders and Marcus), Laurence (UK) and Andrew (US, Arizona) also went to lunch with us today. Well actually Andrew had visited this nice place just "three blocks away" the night before or something like that. As it turns out his spatial judgement skills had been somewhat hindered by the amount of alcohol ingested that night. It was about 8-10 blocks away. Oh well we had a nice lunch anyway.

Jon with his laptop out, Marcus, Laurence, Anders and Andrew.

Research papers
Design fragments among others seemed like novel concept to me at first, but isn't that a lot of work for the framework provider? On top of writing good examples I mean. Well of course it is, but enough for the gain? Maybe I just know too little about it or the examples were too simple.

JTL seemed good in the simple cases and not so easy (which was the goal) in the more complicated cases. Other research paper presentations I listened to were about aspect oriented implementation details improvements (times two) and how to provide uniform dynamic proxies for java (not just for interfaces).

In the evening there was the tribute to John Vlissides, the renowned co-author of "Design patterns". A dessert reception sponsored by IBM. It was followed by a panel discussion involving the other members of the Gang of Four.