Another related OO Python ?

[RJB]

On Feb 16, 6:21 am, Eric Brunel <eric.bru... at pragmadev.nospam.com>
wrote:
> In article
> <6849fd3f-5116-4b35-b274-dc76ae39f... at a11g2000pro.googlegroups.com>,
>
>
>
>
>
>  RJB <rbott... at csusb.edu> wrote:
> > On Feb 16, 12:48 am, Eric Brunel <eric.bru... at pragmadev.nospam.com>
> > wrote:
> > > In article <iJg5p.198317$Mg5.147... at en-nntp-06.dc1.easynews.com>,
> > >  Doug Epling <wde... at mikrotec.com> wrote:
>
> > > > hey, does anyone find the UML useful during Python development of larger
> > > > projects?
>
> > > Well, UML being very Java/C++ oriented, I found out that Python idioms
> > > were really difficult to represent in the diagrams. So I'm using it to a
> > > very small extent and for documentation only, just to give an idea about
> > > how classes are organized. For the rest, and IMHO, it's really too
> > > impractical to be of any use.
>
> > Which of the 13 diagrams have tried and rejected?-)
>
> Diagrams that aren't too bound to the language like e.g the deployment
> diagram can still be used, of course. I was mainly talking about the
> class diagram, which is still the central point of a model. But I even
> found sequence diagrams quite hard to write for Python, unless they are
> very simplistic ones.

Yes. Especially if you draw interactions with a tool and make them
tidy.  I believe in very rough and rapid sketches in pencil or chalk.

Complex code either needs rethinking or in the worst case an activity
diagram, IMHO.

Of course doing the diagrams by hand after the code is OK .... How do
you justify the time spent doing it?