Quest in Time

Tuesday, January 25, 2005

To Tree or not to Tree

I feature that I believe is common to all VS2005 DSL's that I dislike is that they are inherently treelike, and have treelike visualisations. If this is correct then its a key difference from UML (all all the other OOAD modeling languages I've used).

Most of the OOAD methods and techniques I use (CRC, robustness, separation of concern) have always been done with modeling languages that support flat graphs or nodes (classes) and arcs (relationships).

Can these work with trees (acyclic directed graphs?)? If not then we will need a whole new set of techniques.

Most of the above techniques have been around for years, do Microsoft or others, already have a set of tree based techniques, that we can just learn? If not then method maturity may become an issue, i.e. We have lovely shiny new DSL's, but don't have the experience and hindsight to use them effectively.

If its possible to map these techniques successfully onto this world of trees then it may not be such a problem.