Consider these snippets, each run with

>>> import pygst; pygst.require ("0.10"); import gst

before:

>>> gst.Controller ("spam")
Segmentation fault

>>> gst.Controller (gst.element_factory_make ("volume"), None)
Traceback (most recent call last):

+ Trace 72984

>>> gst.Controller (gst.element_factory_make ("volume"), "volume", -1)
Segmentation fault

>>> c = gst.Controller (gst.element_factory_make ("volume"), "volume")
>>> c.remove_properties (None)
Segmentation fault


After attached patch is applied, these come out as:

>>> gst.Controller ("spam")
Traceback (most recent call last):
  File "<stdin>", line 1, in ?
TypeError: argument 1 must be gobject.GObject, not str

>>> gst.Controller (gst.element_factory_make ("volume"), None)
Traceback (most recent call last):
  File "<stdin>", line 1, in ?
TypeError: expected string or Unicode object, NoneType found

>>> gst.Controller (gst.element_factory_make ("volume"), "volume", -1)
Traceback (most recent call last):
  File "<stdin>", line 1, in ?
TypeError: expected string or Unicode object, int found

>>> c = gst.Controller (gst.element_factory_make ("volume"), "volume")
>>> c.remove_properties (None)
Traceback (most recent call last):
  File "<stdin>", line 1, in ?
TypeError: expected string or Unicode object, NoneType found

A lot nicer, isn't it?  The patch also fixes leaking the GLists (though their data is _not_ leaked, the string pointer destinations are owned by python).

There are still more ways to crash, I will provide more patches later and close the bug once I'm satisfied :-)