Wednesday, October 25, 2006

Country Breakfast

The morning started with an invited speaker: Brenda Laurel who was speaking of how to ubiquitous distributed computing to get in more contact with nature. Interesting quotes where Look for the Void, and Look where nobody is doing anything. Interesting thoughts!

After the break, Dick Gabriel and Ron Goldman, talked about Conscientious Software. Dick Gabriel is a big man in the OOPSLA community and he wrote an article many year ago called Lisp: Good News, Bad News, How to Win Big, which I just happened to read on the way over here! In it he explained a philosophy called Worse is Better. Very interesting!

The presentation was a mix of biology, music, city planning and design. A very thought provocative presentation that succeeded in making me think about the inherent problems with complex programming. The paper that the presentation is related to is about software that need to be aware of it’s environment and of itself.

The afternoon I started with Framework and Tools which contained three research papers:

A Framework for Implementing Pluggable Type Systems was about creating a new annotation based type systems for Java that would allow you to specify more specific types such as @NonNull and @ReadOnly

Design Fragments Make Using Frameworks Easier. The reason for code fragments is that programmers copy and paste code and many things may go wrong. The code may be wrong to start out with or the context may not be the same as the one it will be reused in. One good idea that was presented was to have a repository of code fragments index by name and context. However, the idea of letting programmers reuse code fragments by allowing them to bind certain variables in a code fragment does not seem to be a usable way to work to me. I would prefer to solve the same problem with macros and multiple inheritance.

JTL – The Java Tools Language is basically a query language for posing structured queries of code. It is to be used by tools such as IDE’s and as an aspect query language to find join-points. I don’t think it will be used in practice.

In the late afternoon I watched a panel about Ultra Large Scale Systems with, Steven Fraser, Gregor Kiczales, Ricardo Lopez, Peter G. Neumann, Linda Northrop, Martin Rinard, Douglas Schmidt, and Kevin Sullivan.

The solution to ULS is not about formal specification but more about creating an environment that will allow heterogeneous systems to communicate. Conways Law states Any piece of software reflects the organizational structure that produced it. Two software modules A and B cannot interface correctly with each other unless the designer and implementer of A communicates with the designer and implementer of B. and is a real problem in ULS design since there are too many people involved to allow efficient communication. There also seems to be a need for new abstractions but nobody knows what this abstraction should be. Overall, there seems to be a huge problem in this area of computing.

No comments: