Having unwillingly extensively experienced OMT/Rational Rose, this article was rather appreciated. It managed to address my one major problem with OMT/RR - "What the @#$(*&% is this?" (The other, still unaddressed concern is "#$(*&%! It crashed for the umpteenth time and I lost everything; this is worse than diglog!"

I am still trying to decide if Rumbaugh diagrams are the way to go. Although they are certainly the standard, I don't know if that is the best thing for people. Neil's suggestions to the regular system are certainly an improvement (especially on the UI end) because I find that students have a hard time dealing with OMT/Rational Rose because they don't understand the symbols. Making simple improvements on that are definitely to the advantage of the user.

I do have a question - what would be a _good_ way of designing a program? Ignore the fact that UML is the standard (and that CRC cards only help UML), what are other possibilities? I feel extremely limited by UML but I have no idea what I would want in a system to design OO code.

If someone hasn't already said it I will, "All standards are set in stone precisely because stone may be smashed to pieces with a simple hammer."

Having said this, I do agree that it is a good idea to question if UML is the best way to illustrate design. Probably not, but that is just a side effect of evolution. Nothing now is as best as it can ever become. Period. But I do think it is a good way. I'd still love to personally tinker with Tinker Toys (pun intended) to see how hard 3D design is a) to implement and b) to understand. I would imagine that a large system might be much less practical, but who knows. Actually, Tinker Toys are way to rigid, you would need a better medium, oh well.

Probably, like most things, there isn't any one all-comprehensive way that design is best expressed. For most people it probably takes a multitudinous amount of ways to think about and implement design. However in order for people to communicate there needs to be a standard, hence language. I am sure you could argue, at some level, that every language isn't 100% great yet we have them. UML is probably the same way.

So what to do? Try and realize your (everyone's!) inner conceptualizations about design and make them concrete. Maybe you will find that it is a useful thing to do!