[physfs] Wishlist

Ryan C. Gordon icculus at icculus.org
Mon Feb 1 19:27:26 EST 2010

> SWIG sounds promising. Has anybody experience with it, especially with C++
> objects?

SWIG can do C++, but only in limited ways. For instance, classes can be 
done; templates: not really. You can handle operator and method 
overloading, but it requires some tapdancing.

Since PhysicsFS's public API is completely C-callable, it's not a 
problem here.

SWIG doesn't, however, support callbacks. So you can't use 
PHYSFS_enumerateFilesCallback() and have PhysicsFS jump back into, say, 
a Python function. Hopefully they'll add this at some point.

That being said, a first shot at Perl bindings using SWIG are now in the 
  default branch. I'm still working on it. There's a basic definition 
language available to SWIG so we can make some things more Perl-like 
(etc) than a simple wrapper over the PhysicsFS API.

Jury's still out on SWIG. This is all experimental at this point, so I 
might replace it with something else, still. I could definitely see lots 
of scenarios where SWIG would be awesome, though.


More information about the physfs mailing list