Re: Introduction

[Cyrille Chepelov <cyrille chepelov org>]

Le Mon, May 13, 2002, à 05:22:05AM -0700, James Michael DuPont a écrit:
> Hello Dia Developers,
> My name is James Michael DuPont, and I am working on
> the extraction of information from the gcc via the
> parse trees.
> I would be interested in using DIA as a display tool
> for compiler graphs and UML diagrams.
>
> Also I would like to use a modifed version of VCG tool
> to do graph layout, and emacs for source browsing.
>
> Any Ideas?

Hmmm. Assuming you can retrieve the class information from gcc (have you
done it ? I've done some stuff in that area, you may want to have a quick
look at the tarball at http://www.chepelov.org/cyrille/cplusplus), all you
have next is to build an XML file conforming to the dia DTD. You'll probably
need to zless a couple dia files to figure out the exact format used by
dia's UML stuff.

One problem you'll have is figuring out the size of the "UML class" objects
so that you can get VCG to lay out things. Problem, to do that you'll need
to either load the not-yet-laid-out dia file into dia (requires X) so that
dia computes the bounding boxes, or retrieve the font information and
compute the bounding boxes yourself (very likely to require X as well). Or,
you could make some assumptions on the fonts used and the statistical text
sizes, and make educated guesses on the bounding boxes.

Assuming you can make something out of the GCC parse tree information, and
you know how to talk to VCG to extract out of it the information you want,
building the dia file should be a piece of cake !

All in all, this looks like a very fun project, and it would certainly help
giving life to a wonderful tool (have it cooperate with doxygen to make
nicer collab graphs, and you'd make my day big time !)

        -- Cyrille

-- 
Grumpf.