I have got a lot of ideas for this term: Spring and its new style of handling context through dependency injection and inversion of control needs to be investigated. We will be more critical of patterns as well: show that behind many iterator/visitor patterns is simply the need for higher-order function support which is still missing in many OO languages. Prototype concepts need to be known better. We will discuss "Emancipating Instances from the Tyranny of Classes in Information Modeling by JEFFREY PARSONS and YAIR WAND in the context of data access patterns. The observer pattern can be used to show the evolution of a simple (and not really implementable pattern) into high-speed event mechanisms and enterprise message bus technology. Java NIO is still not mainstream knowledge of many developers and we need to show the differences compared to the old mechanism. Can design patterns form an executable language? see the discussion at Design Patterns as Higher-Order Datatype-Generic Programs. Jeremy Gibbons (2006).

The huge number of participants again forced a change in style and content. The hopes for fewer participants have not come true as we are right now in the middle of the diplom/bachelor/master change. This cannot be helped and we will try our best. This is what I would like to do:

For a first participants which pick a pattern for presentation will have to show its use in some software project - preferably from Sourceforge. This way students will learn how to deal with source code AND concentrate on the architectural solution that is provided through the specific pattern..

A next step would be to focus primarily on the problem aspect and let the students decide which patterns they would use to solve a problem. This would mean we need not a list of patterns but a list of specific problems, e.g. when and how to use "context" or when and how to achieve decoupling between objects, components etc. At the end we would get another catalog. This time not on patterns but on complete architectural solutions for standard problems. One could perhaps call this a catalog of applied pattern systems.

Another idea is to focus on eclipse only and uncover all the patterns used in its architecture.

And finally a list of patterns I would like to see explored better:

I guess I am going to change the seminar slightly. After the usual first phase of introducing a number of basic patterns I'd like to dig into eclipse source to show the patterns in action. How is extensibility realized? How does observer work in eclipse? etc. Base will be the new book by Erich Gamma et.al. on Eclipse. It does not provide much more than the web information on eclipse but it structures and brings order to the whole. The section on patterns is especially nice.

I also like the "rule" approach in this book and would like to discuss it with the students.

Other things I'd like to touch this term:

I don't want to promise too much at that time but I will try to get Claus Gittinger, on of the worlds leading smalltalk experts who happens to be an old friend of mine to join us for a session on patterns and architectures in other languages

A discussion of architecture problems and how they can be solved.