Original commit message from CVS:
* testsuite/Makefile.am:
* testsuite/runtests.py:
* testsuite/cleanup.py:
Added cleanup file to initialize registry before running checks.
Added G_DEBUG=gc-friendly to valgrinding
Only test test_*.py files
Original commit message from CVS:
* pygst.py.in:
Don't import non-used modules
* testsuite/Makefile.am:
Heavy valgrinding por favor !
(%.valgrind): with always-malloc for GSlice
Original commit message from CVS:
* Makefile.am:
make check-valgrind calls make valgrind in the testsuite directory.
* testsuite/Makefile.am:
prepend $(top_srcdir) to PYTHONPATH for all tests so it uses the built
library and not the installed one.
* testsuite/python.supp:
Added suppressions discovered on x86/gentoo.
Original commit message from CVS:
* testsuite/test_message.py:
GstBus is flushing in NULL, so we need to set the pipeline to READY in
order to receive the messages in the bus watch.
Original commit message from CVS:
* testsuite/test_ghostpad.py:
* testsuite/test_pad.py:
Another attempt at making the tests deterministic on the buildbots
Original commit message from CVS:
* testsuite/test_ghostpad.py:
Add while loop in teardown to wait for the pipeline state
to hit NULL. Hopefully this will ensure the refcount has always hit 1.
Original commit message from CVS:
* gst/gst.defs:
When using gst.Element.send_event() and gst.Pad.send_event() we keep
a refcount on the sent event.
* testsuite/test_event.py:
Uncomment test to prove above fix
Original commit message from CVS:
* testsuite/test_event.py:
add a test case for autoplugging behaviour:
create a source, connect probes, store new-segment event,
add element in buffer probe callback, and forward event
Currently fails due to refcounting on the stored new-segment
event
Original commit message from CVS:
* gst/gststructure.override: (_wrap_gst_structure_set_value):
Properly check for gstvalue data types. This makes it possible to
set fields with fractions.
* testsuite/test_caps.py:
* testsuite/test_struct.py:
Test for above modifications
Original commit message from CVS:
2005-12-05 Andy Wingo <wingo@pobox.com>
* testsuite/test_registry.py (RegistryTest.testPluginList): Look
for staticelements, not gstcoreelements.
Original commit message from CVS:
* configure.ac:
* gst/Makefile.am:
gst-interfaces-0.9.pc is no more, update to new use
* testsuite/test_element.py:
Fix for queue modification
Original commit message from CVS:
* gst/gst.defs:
Keep refcount for the message of gst_bus_post()
* gst/gststructure.override:
Don't free a structure on dealloc (the parent does that)
* testsuite/test_message.py:
Proper testing of bus functionnality with a mainloop
Original commit message from CVS:
* gst/gstelement.override:
* gst/pygstexception.c:
* gst/pygstexception.h:
* testsuite/test_element.py:
add gst.QueryError and use it
* testsuite/test_pad.py:
add some tests that show comparison between two different
Python objects wrapping the same MiniObject
Original commit message from CVS:
* gst/gst.override:
* gst/pygstminiobject.c:
remove the hash table for miniobjects - since we can't get notified
when they get destroyed, we shouldn't be caching pointer mappings
* testsuite/test_pad.py:
update refcount tests because mini objects now have a ref for
each time an object is made for it
Original commit message from CVS:
* gst/gst-types.defs:
* gst/gst.defs:
Update (hopefully) final API modifications
* gst/gstelement.override:
* gst/gstpad.override:
* gst/gstquery.override:
Updated position querying/parsing
Removed useless debug line
* testsuite/test_element.py:
Updated test for new API
Original commit message from CVS:
* gst/gst.defs:
* gst/gst-types.defs:
Updated API
* gst/gstelement.override:
_wrap_gst_element_get_state: made wrapper behave more like codegen-ed
functions.
It now only takes an int or long as optional timeout argument, and if
there's no argument it will use GST_CLOCK_TIME_NONE
* testsuite/test_bin.py:
* testsuite/test_event.py:
* testsuite/test_ghostpad.py:
* testsuite/test_pipeline.py:
Updated testsuites for element.get_state() modifications
Added event parsing checks
Commented out test_ghostpad.PipeTest.testProbedLink til I find out why
"while self._handoffs < 10: pass" doesn't exit although it reaches
values way over 10
Original commit message from CVS:
* gst/gst.defs:
* gst/gstmessage.override:
update for latest API changes
* gst/gstelement.override:
use GstClockTime for get_state
* testsuite/test_pad.py:
add logging
Original commit message from CVS:
* gst/gst.defs:
Update for new API
* gst/gst.override:
Added unreffing for gst.Registry.get_plugin_list and .get_feature_list
* testsuite/common.py:
removed crude hack for gst.Registry and gst.Plugin
* testsuite/test_interface.py:
* testsuite/test_caps.py:
* testsuite/test_struct.py:
* testsuite/test_xml.py:
converted to using common's TestCase class
* testsuite/test_event.py:
Enabled/fixed previously non working test
* testsuite/test_registry.py:
Now uses common's TestCase class,
added special case setUp()
Original commit message from CVS:
* gst/gst-types.defs:
* gst/gst.defs:
Updated to new API
* gst/gstbus.override:
bus.add_watch() now uses gst_bus_add_watch_full() using
pyg_destroy_notify as the DestroyNotify function, thus enabling
proper python garbage collection.
* testsuite/common.py:
Let's check refcounting of ALL gst.Object
* testsuite/test_bin.py:
Added debug
* testsuite/test_bus.py:
Added "message" signal version of test
Original commit message from CVS:
2005-10-12 Julien MOUTTE <julien@moutte.net>
* gst/gst-argtypes.c:
* gst/pygstminiobject.c:
* testsuite/testhelpermodule.c: Reordering header inclusion
order to
fix build on my machine.
Original commit message from CVS:
* gst/gst.defs:
* gst/gstmessage.override:
fix for pending state being added to state message
* testsuite/test_buffer.py:
* testsuite/test_pipeline.py:
update
Original commit message from CVS:
* gst/pygstiterator.c:
Unref the return value, since the iterator refs them.
* testsuite/test_iterator.py:
Added more test cases, re-enabled gcverify now the bug is fixed
* gst/pygstminiobject.c: (pygstminiobject_register_wrapper),
(pygstminiobject_new), (pygstminiobject_dealloc):
Comments to better track the creation/destruction of PyGstMiniObject
Fixed naming (gst.GstMiniObject => gst.MiniObject)
* testsuite/python.supp:
Updated python specific valgrind suppressions to latest version and
added x86_64 cases
* testsuite/gstpython.supp:
Series of suppressions for errors/leaks not solvable within gst-python
* testsuite/Makefile.am:
Added gstpython.supp
Original commit message from CVS:
* gst/pygstminiobject.c: (pygstminiobject_register_wrapper),
(pygstminiobject_new), (pygstminiobject_dealloc):
Comments to better track the creation/destruction of PyGstMiniObject
Fixed naming (gst.GstMiniObject => gst.MiniObject)
* testsuite/python.supp:
Updated python specific valgrind suppressions to latest version and
added x86_64 cases
* testsuite/gstpython.supp:
Series of suppressions for errors/leaks not solvable within gst-python
* testsuite/Makefile.am:
Added gstpython.supp
Original commit message from CVS:
* configure.ac:
need at least 2.6.3, because that contains the code that
converts signal closure arguments to the correct Gst
types
* gst/gstpad.override:
log name of probe handlers we call
* testsuite/test_pad.py:
fix up tests; use TestCase base class methods
Original commit message from CVS:
2005-10-06 Thomas Vander Stichele <thomas at apestaart dot org>
* testsuite/test_ghostpad.py:
add handoff to count received buffers