[plt-scheme] Help writing non-trivial applications.

From: Neil W. Van Dyke (neil at neilvandyke.org)
Date: Sat May 7 03:11:18 EDT 2005

> Since I've been testing different UML tools, trying to understand the 
> organization of some 620 classes that others built, I'm curious to know 
> your opinion of UML's grievous errors.  I navigate the object map much 
> faster in Emacs than I do in a UML tool.

The first and most bothersome to me was the dropping of graphical
notation for the multiplicity/cardinality of association roles.

I say "dropped" in UML because Booch-Rumbaugh-Jacobson clearly had
started with the static object model notation of Rumbaugh, et al. OMT
rather than Booch's comparable own notation, and missing multiplicity
graphics were perhaps the most glaring change.

In OMT, these were gloriously useful little circle graphics on the
association arcs in OMT that distinguished "exactly-one", "optional",
and "many".  They were a boon to at-a-glance visualization of large
graphs, and I believe they were one of the best visual design decisions
in the OMT languages.

A conspiracy theory -- given that all three of the authors by that time
worked for CASE tools vendor Rational -- would be that Rational's
diagram editor could not cope with the editing constraints imposed by
the circle graphics, and Rational didn't want to rewrite the editor. :)

I tried to get the graphics put back in before UML 1.0, but failed.  I
stopped tracking UML after that, so I can't comment on the mountain of
specifications they seem to have produced since then under the OMG

Posted on the users mailing list.