On Sat, Mar 31, 2012 at 2:34 PM, Mark Fortner <[email protected]> wrote:

> Does the process above roughly fit the way everyone else is using ArgoUML at
> work?  Has anyone had any problems getting other people to use ArgoUML at
> work?  If so, what get's in the way?  How do people use ArgoUML in teams?
>  Do you do design reviews, if so, how do you use ArgoUML to make that
> happen?  Do you generate code from ArgoUML models?  If not, why not?

I'd argue that the developers are the least interesting community to
ask this question of.  In my mind, many of ArgoUML's shortcomings can
be traced to lack of engagement with real world users.

Some of the things I use (or have used) ArgoUML for include:

- to sketch little design vignettes to share with people (a handful of
classes on a single diagram)
- to look at UML models created by other people (often standards
groups, usually using a different tool, which can create interchange
problems)
- to visualize and understand new, large code bases using reverse
engineering (not very successfully usually due to some of the problems
others have mentioned)
- as an input tool for MDA using AndroMDA (mainly to get a lot of the
db table generation cruft done like Andreas)

I would never attempt to keep code and UML in sync using ArgoUML.
That's an awful lot of effort for very little gain.  I *might* try to
do it if I had really excellent round-tripping support available to
me, but probably not even then.

Thinking of UML as the "design" is a fallacy in my mind.  The design
is an abstract thing which lives in the shared understanding of the
team members.  To the degree that the team members haven't converged
on a shared design, there are N designs (or perhaps a single design
with fuzzy edges).  The UML, the design documents, the code, the white
board drawings are all *representations* of the design, with varying
degrees of fidelity, but they are not *the design*.

I don't think the industry has yet to come up with a way to capture
designs which is lightweight enough to use on a consistent basis, but
also preserves all the important aspects of the design.  You either
end up investing huge amounts of time/money in design documentation
which a) usually isn't used and b) when it is used is *still* missing
some important aspect of the design or you end up going the "agile"
approach and maintaining the design information using social
constructs (supported by code & tests), but leaving yourself at risk
for losing key design information when staffing changes.

Tom

p.s. It's been a while since I played with it, but a local (to me)
company that's trying to address some of these challenges is
Architexa. http://www.architexa.com/  They've got an online
collaboration site for open source projects at
http://www.codemaps.org/ which demonstrates some of the
collaborative/review aspects.  Not a plug for a competitor - just
something interesting to look at for people interested in this space.

------------------------------------------------------
http://argouml.tigris.org/ds/viewMessage.do?dsForumId=450&dsMessageId=2943771

To unsubscribe from this discussion, e-mail: 
[[email protected]].
To be allowed to post to the list contact the mailing list moderator, email: 
[[email protected]]

Reply via email to