> -----Original Message-----
> From: segg [mailto:[EMAIL PROTECTED]]
> Sent: maandag 7 februari 2000 16:36
> To: [EMAIL PROTECTED]
> Subject: Re: UML diagrams -> source code conversion
> 
> 
> [EMAIL PROTECTED] wrote:
> > 
> > On Mon, 7 Feb 2000, Stefanus Du Toit wrote:
> > 
> > > Hi,
> > >
> > > I have been using dia for a while now to create UML 
> diagrams for various
> > > C++ projects of mine - I must say, it's very nice and has 
> matured very
> > > well. Thanks!
> > >
> > > I'm wondering whether there is any interest in writing 
> some kind of UML ->
> > > source code converter (I would be very interested in 
> getting something
> > > like that going for C++/Python, even Perl) that supports 
> the XML output
> > > generated by Dia.
> > >
> > > It shouldn't be all that hard, thanks to the output being 
> in XML. If there
> > > is such an effort already started, I would appreciate any 
> pointers to
> > > it. Otherwise I'll probably start a dia->c++ conversion 
> tool of my own.
> > >
> > 
> > There is a standard XML DTD for UML. It is called XMI, and it is
> >   intended to enable the separation of the code generator from the
> >   diagraming utility. (e.g., use dia to draw the UML diagram, dump
> >   it in XMI format, grab a separate XMI to C++ converter, 
> and use that to
> >   generate C++.)
> 
> And source C++ reverse engineering to UML.
> Needs, choose a repositery for the code generated, select C 
> and H directories.
> Call the reverse C++.  Allow adding more source to UML.
> We need also to specify how many levels deep you want on a drawing.
> This means we end up with many drawings.
> 
> Consequence, we need a data base to know how/what to update 
> the other diagrams
> after a modification.  That is nice, I see a big decision here.

It seems like you want to rebuild DIA to a full blown CASE tool! This is not an easy 
thing to do. I've been working (well, thinking mainly ;-) about how to get it right, 
but haven't produced anything useful yet, except for a nice setup for a UML repository 
(generated from the original UML metamodel).

> We need a development tree(cvs branch) for this, is it possible.
> Do we have any "open reverse C++" tools already available.

None that I know of.

At your service,

Arjan

Reply via email to