Python and Visual Programming?? Ideas for a complete python "IDE"

David LeBlanc whisper at oz.nospamnet
Mon May 14 12:05:07 EDT 2001

In article <3AF945DB.7B90AA6D at>, Nicolas.Chauvat at 
> Dia doesn't even come close to a modelling tool. It is a graph drawing
> tool that provides ready-to-use UML classes-look-alike boxes. but that's
> about it. What's the difference you may ask? Well try doing this with
> Dia: create a class, then another that inherits from the first. Then
> create another class diagram with the first class (just copy it over ).
> Now let's say you copy over the second one. Hey, where did the
> inheritence link go? As you've figured out, it never was an inheritence
> relationship, it was a 'straight link from that first box to that other
> box'. If you really want to use modelling tools, I suppose you don't
> want to use Dia ;-)
I don't disagree that Dia (done in awhile?) isn't a true modelling tool 
as it has no information about the relationships between the "blobs" it 
puts on the screen, except possibly to rubber-band connections when said 
blobs are moved... but it's a start.

"There is also a program available for Dia called Dia2code, which will 
convert Dia UML diagrams into C++ or Java code for you."

Alas, python isn't mentioned - but it could be! :)

> But the world is not so harsh as open source modelling tools do exist.
> Try for start and you'll get most of the UML diagrams as
> only sequence diagrams are missing (Dia would only provide you with
> boxes for the class diagrams).

Good stuff Argo - only mildy hindered by being in java...

> Now if you're interested in round-triping from model to python code, and
> back to model, you'd better use XMI. XMI is a standard XML serialization
> format for UML. A lot of UML tools can write and read XMI. Hence a nice
> solution sounds to convert XMI to python code. And back.
> will help you on the way back as it takes
> python code and serializes the parse tree to XML, then turn it into a
> text-formatted model. It's not XMI yet, but that's the goal.
> Hope this helps,


Dave LeBlanc

More information about the Python-list mailing list