>When I get specs from my system admin, if I think it's going to take anywhere's close to a week I'll do up a UML diagram. Only takes a few hours to get a working and logical diagram. Not only does it provide me with a great reference but it's invaluable as a reference for those who aren't immediatly within the development process. That could be someone who drops into the project 2 days in or who does some tinkering with it years later.


I'll admit that a lot of factors would go into the decision. And, as a documentation tool, it is a great idea. But I think that the earlier point is valid -- the entire science of UML is geared so much toward larger apps that while it is never a drawback for a small app, it can create such overhead that the time required (and price if you are a freelancer) can defeat the purpose.

If I was building a tool shed, I'd love to have an entire set of very precise blueprints, and put in the strongest foundation possible. Then again, I might not have the cash at the moment, and might really really really need a basic shed. In that case, a quicker approach might make more sense....

... or maybe not! Again, so many factors go into these decisions.