Re: UML & Ada

["Marin David Condic" <dont.bother.mcondic.auntie.spam@[acm.org>]

Thanks. I'll try to take a peek at CAIM - do you have any links?

I'd like to see a modeling language that did a couple of things: One would
be to have diagram elements that represent Ada constructs (packages, tasks,
subprograms, etc.) with varying levels of "visibility" (e.g. "show me this
package and its subprograms, but not its types and objects...") to which you
can attach matching code (either partially generated or all manual) so you
can punch a button and get the code from the diagrams. This would be useful
for describing the program starting at a fairly high level and iterating
down to a final product. The important thing being that you have a method of
representing anything that can be built in Ada. (Booch diagrams from several
years ago did a fairly nice job, but the last tool I used that did this was
Adagen - and while I found it to fit the niche reasonably well, they had
problems with really buggy software & poor support. I believe they went
casters-up - or maybe got bought up, but I've not heard anything about the
tool in years.)

The other thing would be to have some ability to describe a program at a
higher level of abstraction - I've found interaction diagrams to be useful
and I can see having something that supports object oriented design (that
breaks down into actual Ada components) or other methodologies. At a lower
level, I still find flowcharts to be occasionally useful - especially if you
can generate code from it.

I've found UML to be less than intuitively obvious to the casual observer.
Its been a long time since I've been out searching for a methodology tool,
so maybe there is something out there a little more to my liking. I may have
to once again dive in and look more seriously.

MDC
--
Marin David Condic
Senior Software Engineer
Pace Micro Technology Americas    www.pacemicro.com
Enabling the digital revolution
e-Mail:    marin.condic@pacemicro.com
Web:      http://www.mcondic.com/


"Jeffrey Carter" <jrcarter@acm.org> wrote in message
news:3BF9EB02.D9946B9B@acm.org...
> Marin David Condic wrote:
>
> I'm not very fond of UML, as it seems you need a large number of
> diagrams, each of which contributes a couple of bits to the total
> description. I much prefer something like CAIM, in which a few diagrams
> each contribute a great deal.
>
> It also seems as if UML works best for business applications in C++, so
> of course there would be no concept of tasking.
>
> That said, however, I think a modeling notation should not represent
> language-specific implementation features. The initial default
> assumption should be that all objects are active, and that can then be
> modified as you work towards an implementation.
>