Original commit message from CVS:
* gst/gstobject.override:
Release the GIL when calling gst_object_get_path_string() since it can
cause deadlocks with new pygobject behaviour.
Original commit message from CVS:
* configure.ac:
fix use of PKG_CHECK_MODULES
check for a pygobject of at least 2.11 for a value_from_pyobject fix
* gst/gstobject.override:
use it
Original commit message from CVS:
* gst/gst.defs:
Add declaration of gst_object_set_property so we can use our MT-safe
version of set_property().
* gst/gstobject.override:
Implement a MT-safe version of g_object_set_property for GstObject.
The problem is that currently g_object_set_property is called in
pygobject with the GIL lock taken. This can cause deadlocks.
Remove this hack once bug #395048 is fixed in pygobject and we depend on
the fixed version.
Thanks to Lord Wingo of the "realm.py haters club" for proposing the
idea.
Original commit message from CVS:
Reviewed by Edward Hervey <edward@fluendo.com>
* gst/gstobject.override: (_wrap_gst_object_tp_traverse):
Py_VISIT() might make code cleaner... but it doesn't work with
python < 2.4
Closes bug #318692
Original commit message from CVS:
* gst/gstobject.override:
don't try and unref objects that are already finalizing
* gst/gstpad.override:
fix up the set_chainfunc method
* testsuite/common.py:
* testsuite/test_pad.py:
add linked/unlinked tests with no/true/false bufferprobes
Original commit message from CVS:
* gst/gstbuffer.override:
a buffer is not a boxed
* gst/gstobject.override:
use Py_VISIT for cleaner and more pythonic code
do not visit GstObjects that still have a parent
