I'm experimenting with displaying 3D models in a PyGtk DrawingArea. To
get (re)display fast enough, I "compile" the model into a set of line
segments that can be drawn witk GtkDrawingArea.draw_segments. The
problem with this approach is that tens of thousands of line segments,
represented as a list of quadruples of integers, consume a LOT of
memory. It also takes a lot of time to build up the structure, and
then take it apart again when the low-level pygtk code calls
gdk_draw_segments. To get around this problem, I'd like to keep the
line segment list as an array of GtkSegments and do the
gdk_draw_segment call from C.
However, there does not seem to be a "clean" way for a module written
in C to get hold of (for example) the GdkWindow* for a drawing area
created with pygtk. Of course, I could use PyGdkWindow_Get, but this
would mean that my C code had to include header files from pygtk, and
possibly also link with parts of pygtk, which feels a little
uncomfortable.
One way to work around this would be to add a method to the PyGdk_Foo
types, that returns the underlying GtkFoo* wrapped in a PyCObject.
That way, it would be possible for Python code to pass the low-level C
objects into a module written in C that performs direct gdk calls, and
still have the C code independent of the inner workings of pygtk. I
have implemented this in my own copy of PyGtk; if anyone else thinks
it's a good idea, I'll try to extract a patch from among all the
~-files sitting around my ~/src/ directory. :-)
Another wild idea on the same theme (I haven't tested this one,
though) is to give the PyGtk_Object class a python-callable
constructor that takes a CObject as argument. This could be used to
interface third-party Gtk widgets into pygtk, without having to modify
pygtk itself.
-- Ture
To unsubscribe: echo "unsubscribe" | mail [EMAIL PROTECTED]
