[PD-dev] file library WAS: [folder_list]

[Mathieu Bouchard]

On Sat, 1 Apr 2006, David Plans Casal wrote:
> On 1 Apr 2006, at 01:21, Mathieu Bouchard wrote:
> > We're using SWIG to connect LTI to Ruby so that it becomes easier to
> > connect to Pd, because of GridFlow.
> LTI?

  http://ltilib.sourceforge.net/doc/homepage/index.shtml

> Don't need to bypass it, and don't necessarily think I will, I'm just
> trying to see if the SWIG dependency would be less troublesome, as
> gridflow is a little hard to compile on osx,

Well, there have been some problems recently. I rewrote the Makefile so
that it doesn't use mkmf.rb (a Makefile generator) because mkmf.rb made
things more difficult. GridFlow 0.8.2 doesn't compile on OSX, but 0.8.3
already does.

> and almost impossible in windows, etc.

If no-one ever talks to me about their attempts to get GridFlow to work on
win32 I can't even start helping them!!!

> But then I think I may be swapping one set of problems by another, and
> then again, I could be helping to debug gridflow compilation instead of
> checking SWIG... ;-)

It would be interesting to have a Pd module for SWIG, but I seriously
doubt that we could build a reasonable set of typemaps that covers most
cases. I mean I even doubt something could map Ruby to Pd automatically: 
instead I either write wrappers myself (they usually aren't long to write 
anyway) or I write a wrapper generator for a specific set of classes.

If you have a class in C++, how do you determine whether it should be 
mapped to a Pd class, a family of Pd classes, and/or a Pd atomtype? This 
is a general question...

 _ _ __ ___ _____ ________ _____________ _____________________ ...
| Mathieu Bouchard - tél:+1.514.383.3801 - http://artengine.ca/matju
| Freelance Digital Arts Engineer, Montréal QC Canada