UML Bloat

That Unified Modeling Language (UML) is flawed by feature-bloat and design-by-committee should be obvious to anybody who has had the misfortune of developing software in a large, process-run organization. UML 1.0, which showed early promise as a software modelling language, passed through the collective hands of a committee and emerged smelling like week-old roadkill. I have a strong suspicion that the continued discussion of UML 2.0's merits can be largely attributed to the fact that most of its proponents haven't actually had to use UML in any significant way.

Still, I was surprised to read James Rumbaugh's comments in Masterminds of Programming on UML, given that Rumbaugh is one of the UML's original developers. From p. 342:

What process would you use to simplify the language (UML)?

James: I doubt it can be simplified through a standards process, such as the OMG process that produced UML 2.0. There are too many competing interests all trying to stuff their own ideas into the pie. The problem is that standardization processes place too little emphasis on consistency, simplicity, and uniform style; instead the overemphasize excessive content. In fact, I'm not a big believer in standardization bodies; they tend to produce overstuffed products lacking in elegance and usability. I was reluctant to get involved in them in the first place; my fears about their negative effects have been realized.


Suppose that I am skeptical about UML. How would you persuade me?

James: I'm skeptical about a lot of it myself. I think it has gotten terribly bloated by too many cooks in the OMG kitchen. Also people have tried to promote it as the answer to all things for all people. The whole computing field has a tendency to hype any new development beyond reason. There is also a tendency to look for a single solution for all problems. Life and computing are too complicated for simple solutions.

James had a few choice things to say about patterns, too, but that's another topic.

(oh what the heck, here's what he says on p. 343): In one of your first answers, you said that you think the patterns movement did some great things, but they had too narrow a view of patterns. Can you expand on that?

James: At a workshop I once attended that included some of the Hillsdale Group, it came out that they had this very narrow, almost quasi-religious view of what patterns were, and they really didn't want to expand it. They were very protective of their official view of patterns and this sort of worship of the architect Alexander. They viewed patterns as this very specific thing, and I think patterns can be applied at many different levels.


UML Bloat (last edited 2014-03-24 16:38:43 by RobertYu)