[BangPypers] UML <=> Python...and other things...

[Banibrata Dutta]

On Sun, Mar 8, 2009 at 3:25 AM, Vishal <vsapre80 at gmail.com> wrote:

> Hi,
> Does anyone know of a tool that can produce Python code from UML diagrams?
>

Never used personally, but found few moments back -- 'Gaphor' (
http://gaphor.devjavu.com/) <http://gaphor.devjavu.com/>

I know that BOUML is superfast and is quite extensible with possibility of
writing it's plugins using a well documented API (mixed C/Python plugin
possible). I've used BOUML (but for C++ projects), and it is one of _the_
fastest babes around.


> Is something of this sort being added to NetBeans?
>

Impossible is nothing ;-)

Or is it that the ease of programming in Python acts as a deterrent in the
> way of having to create something in UML and then covert that model into
> Python.
>

Forward engineering should not be a challenge, but any programming language
that has too many ways of representing the same idea, and where the OO
paradigm can be functionally represented in multiple (too many) form,
reverse-engineering (not round-trip, simple reverse) presents some
challenges (not unsurmountable though). Once the forward and reverse engg
parts are individually nailed, the full round-trip thing becomes possible,
though some tools _by design_ choose to work in completely context-free
manner, and make round-trip jobs difficult. My knowledge of Python isn't
good enough to say whether Python is such a language, however given that
Gaphor exists, this (forward engg) is a solved problem.

>
> it would be good to have a tool of that sort (Python to UML and UML to
> Python)?
>
> Also, is there a way to find Python call graphs (something like Doxygen),
> but not just the typical static code structure...instead something that can
> tell execution paths while a certain function is called.
> I came across 'pycallgraph'. Its good. except two things, its slow for big
> projects, and it goes all the way into tracing every single call...that
> means if I am using a COM library underneath, it traces that as well. What I
> was interested in is figuring out only part of the trace...may be specifying
> exclusions in the trace tree.
>

I am guessing that it does dynamic (run-time) analysis, and unlike static
analysis tools, the depth control can only be "simulated", i.e. in reality
full depth traversal happens, hence you find it slow.

-- 
regards,
Banibrata
http://www.linkedin.com/in/bdutta
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/bangpypers/attachments/20090308/4e1155aa/attachment.htm>