[M3devel] Modernising m3-ui?

[Dragiša Durić]

What SWiG does is only tiniest possible piece of work. Modula-3's way is
not about accepting somebody's (especially C-world's) ideas on
modularity/API's/... and mappping variables/procedures/functions/... to
Modula-3 ones. C world  API's are more or less nightmare, my "favourite"
being pthreads which is full of compromises made to get (mainly) Sun's
signature on standard... sqlite3 is another one not too away from....

But, of course, it's everybody's right to spend her time SWiG-ging
around :)... Me, after tons of experience with it, I'll continue to work
mostly by hand. Automated solutions we need aren't in SWiG... To make,
for example, proper Modula-3 objects from Gtk gobjects, we need custom
conversions...

I've already dipped into this with Gtk and it's enlightening... It's
customary in Python/Perl (h2def, gtkdoc) world to use regexps for
parsing.... When I saw it first I smiled, but when I tried two different
tools and got same buggy output, it became more serious :). 

What we need to exploit widely available C libraries is good and
proper .h parser, and then some tools to rewrite them into Modula-3...
What SWiG offers is one posibility, but we need someone proficient with
Modula-3 to join them - first guy who tried it went Haskell some time
ago.

dd

On Fri, 2010-01-22 at 12:05 +0000, Jay K wrote:
> > Main problem with diversity of solutions is multi-platform nature
> of 
>  > Modula-3 - solutions not multi-platform are not likely to be
> accepted... 
> 
>  
> I'm sure we can win here, if we do anything.
>   (I'm not volunteering. The idea of using Swig is good.)
> Just wrap Qt or FLTK or Tk or wxWidgets or such.
> Qt is my preferred.
-- 
Dragiša Durić <dragisha at m3w.org>