Live UML - Explorative Modeling in the small

Looking at the applications programmed in the world, and looking at the programming languages used, the situation seems not positive for Smalltalk. But for me, Smalltalk is not dead. In the opposite: for me it has growing value, although my daily life is determined by C and C++. How can this be ?

If I have to build a new application or a new module, today I use Smalltalk to design the architecture. In general, if one has to do something new, classes must be identified, what they are doing, which data they work on and more. Extensibility (for new requirements) must be kept on focus, and that there will come an interface into life which could be understand and used by others. Here can help Smalltalk a lot: a few objects doing something is written in short time. Every rearrangement and new ideas can be implemented quickly. Because Smalltalk allows to modify code even in debugger, and allows to evaluate any code snipped at any place - the right design grows in a way also evolution works: simple try it. It is growing in the best sense.

At the end, I got a set of objects, a set of methods and communication network (who talks to who), which can carry the basic functionality. From this, it is easy to get UML - for communicating with managers - and the implementation in the destination language. Now, if - as in my case - the destination language is C++, there are some pitfalls of course. But because the runable Smalltalk model has partitioned the problem in static AND dynamic elements, the problems are small and solveable.

In fact, this is xM in the smallest possible scale....

In the future, I plan to create some ready-to-use moduls which are providing elements always used in our company. Radar data parser, or some kind of common displays or GUI elments are examples. Especially when combined with Seaside, this should result in a great Live UML (xM) Framework.

So Smalltalk is not dead, in the opposite, it is THE tool for robust and failure safe development !

Rendering documents - support the semantics

The rendering of documents is only one part of the story. Another goal would be not only to build a tool for creating a document in the sense of set in in a certain form, but also to collect the right content for it. Now it is clear, that it is not possible today to build didactical efficient documents which are optimal for human reading and understanding. But if we not try single steps toward this direction, we will never come closer to the goal

So what can be done today ? Documents are answers to questions. Sounds trivial, but is the jumping point. So maintaining content means maintaining questions. So this could be a simple first step: manage not only content, but also manage the questions relateted to it and manage also the relations. The other thing are terms. Terms are the most basic but important building block of any content. In consequence, building term networks (and visualising them) would be another brick in the wall.

At this time, I am reading something about Husserls Phenomenology, a philosophy which looks (together with other things) how our understanding can come to see the nature of things. May be there are aspects which could be tried in the context of document rendering. And again, computer science will become practical philosophy

Of course, I also try to get an overview ofer things like semantic web.