UML Support for Python [was IDLE development - Call for participation]

[Martin Rand]

On 17 Aug 2000 22:54:16 +0200, cg at gaia.cdg.acriter.nl (Cees de Groot)
wrote:

>Warren Postma <embed at NOSPAM.geocities.com> said:
>>
>>Anybody agree with me? Anybody think I'm nuts?
>>
>Yes; I don't know. 
>
>UML is a nice common language for if you think you need to communicate
>something to a colleague with a diagram. Often, a simple class diagram
>or sequence diagram says more than a thousand words. A complicated
>class diagram says considerably less, and full-fledged round-trip
>engineering thingies tend to generate very complicated class 
>diagrams (like, if you print them, you can paste together 6 pages
>for the diagram and still need a loupe to discern detail...).
>
Fully agree. But I'll use your point about complexity on diagrams to
make another argument *for* diagrams in general. Recently I had to
advise a team of people who were developing a set of reusable software
components for families of plant actuators and instrumentation. They
were struggling to write up a good set of specifications, let alone
implement them. I drew a class diagram of what they were doing and
pinned it up across one end wall of a meeting room, next to the class
diagram for the Java AWT package. The diagrams raised the obvious
question very starkly - why does something with one-tenth of the
complexity of another package require a class diagram six times the
size? But more importantly, they made it easy for everyone to see how
the existing design could be radically restructured through a series
of methodical steps into something much simpler, with a better
distribution of responsibilities. In three hours we covered something
that could have taken days to talk through from the textual specs
they'd produced up to then.

We seem to be getting off topic. I'd suggest moving into
comp.software-eng if others agree.


--
Martin Rand
Highfield Software Ltd
mwr at highfield-software.co.uk
Phone: +44 (0)23 8025 2445
Fax:   +44 (0)23 8025 2445