commit ea992324b8197b2d04ff2849b9ab46f8a04b4ed7
Author: Martin Pitt <martinpitt@gnome.org>
Date: Mon Sep 3 22:04:01 2012 +0200
release 3.3.91
NEWS | 32 ++++++++++++++++++++++++++++++++
1 file changed, 32 insertions(+)
commit 1e1f5b2f2f15547c1f2cbc948d2b764bd0a37c44
Author: Martin Pitt <martinpitt@gnome.org>
Date: Mon Sep 3 21:57:00 2012 +0200
Fix exception test case for Python 2
Regression from commit 77844c5 which did not work with Python 2.
tests/test_everything.py | 12 +++++++-----
1 file changed, 7 insertions(+), 5 deletions(-)
commit 41bb687c058e08b05108b4b2f081cd83d4f93da8
Author: Martin Pitt <martinpitt@gnome.org>
Date: Mon Sep 3 21:16:01 2012 +0200
Bump g-i dependency
Require at least 1.33.9, as we got a couple of bug fixes there which
the tests,
and for some cases the code, depend on. We actually require 1.33.10
for all
tests to succeed, but that hasn't been released yet.
configure.ac | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
commit 77844c571ad0badc189428b93de9f2572051b67e
Author: Martin Pitt <martinpitt@gnome.org>
Date: Mon Sep 3 17:58:38 2012 +0200
Show proper exception when trying to allocate a disguised struct
Instead of a simple "MemoryError" with no details, raise a proper
TypeError with a traceback and an explanation what happened.
https://bugzilla.gnome.org/show_bug.cgi?id=639972
gi/pygi-struct.c | 7 +++++++
tests/test_everything.py | 13 +++++++++++++
2 files changed, 20 insertions(+)
commit 0d099bdb3f4bbd962e5e60b583673d9e6f5673cc
Author: Mark Nauwelaerts <mark.nauwelaerts@collabora.co.uk>
Date: Mon Sep 3 16:47:22 2012 +0200
Support marshalling GParamSpec signal arguments
Fix marshalling GParamSpec arguments from C to Python.
https://bugzilla.gnome.org/show_bug.cgi?id=683099
Co-Authored-By: Martin Pitt <martinpitt@gnome.org>
gi/pygi-argument.c | 10 +++++++++-
tests/test_signal.py | 14 ++++++++++++++
2 files changed, 23 insertions(+), 1 deletion(-)
commit 69fb92c22b3f3d1d5e8c3e14134eee3242fdc5fc
Author: Martin Pitt <martinpitt@gnome.org>
Date: Mon Sep 3 16:06:49 2012 +0200
Add test for a signal that returns a GParamSpec
https://bugzilla.gnome.org/show_bug.cgi?id=683265
tests/test_signal.py | 6 ++++++
tests/testhelpermodule.c | 12 ++++++++++++
2 files changed, 18 insertions(+)
commit a7c524219987fbf37e455a91e4c78d2b9b4db12d
Author: Simon Feltman <s.feltman@gmail.com>
Date: Tue Mar 20 04:33:50 2012 -0700
[API add] Add Signal class for adding and connecting custom signals.
The Signal class provides easy creation of signals and removes the
need for __gsignals__ in client code. The Signal class can also be
used as a decorator for wrapping up the custom closure. As well as
providing a "BoundSignal" when accessed on an instance for making
connections without specifying a signal name string.
Python3 annotations can also be used to supply closure argument and
return types when Signal is used as a decorator. For example:
class Eggs(GObject.GObject):
@GObject.Signal
def spam(self, count:int):
pass
https://bugzilla.gnome.org/show_bug.cgi?id=434924
examples/signal.py | 34 ++++--
gi/_gobject/Makefile.am | 3 +-
gi/_gobject/__init__.py | 5 +
gi/_gobject/signalhelper.py | 251
+++++++++++++++++++++++++++++++++++++++++++
tests/test_signal.py | 208 +++++++++++++++++++++++++++++++++--
5 files changed, 482 insertions(+), 19 deletions(-)
commit 96fa22369fd188465559fc904c7f76e73040e6dd
Author: Martin Pitt <martinpitt@gnome.org>
Date: Mon Sep 3 15:32:12 2012 +0200
Fix pygtkcompat's Gtk.TreeView.insert_column_with_attributes()
We have a proper implementation for insert_column_with_attributes()
now, so
drop pygtkcompat's empty stub for it.
Also improve test case for
Gtk.TreeView.insert_column_with_attributes().
gi/pygtkcompat.py | 6 ------
tests/test_overrides_gtk.py | 16 ++++++++--------
2 files changed, 8 insertions(+), 14 deletions(-)
commit 542cf22c9de9b2094868c4e879b0f24b15c4c012
Author: Marta Maria Casetti <mmcasetti@gmail.com>
Date: Mon Sep 3 13:06:22 2012 +0200
Add override for Gtk.TreeView.insert_column_with_attributes()
https://bugzilla.gnome.org/show_bug.cgi?id=679415
Co-Authored-By: Martin Pitt <martinpitt@gnome.org>
gi/overrides/Gtk.py | 7 +++++++
tests/test_overrides_gtk.py | 38 ++++++++++++++++++++++++++++++++++++++
2 files changed, 45 insertions(+)
commit 1c73e845361e471b1c3a3f17e40e6a6cfa740877
Author: Martin Pitt <martinpitt@gnome.org>
Date: Mon Sep 3 11:09:55 2012 +0200
.gitignore: Add missing built files
.gitignore | 9 +++++++--
1 file changed, 7 insertions(+), 2 deletions(-)
commit 96431f393036a688666dcf67911bf12b9824b264
Author: Martin Pitt <martinpitt@gnome.org>
Date: Mon Sep 3 11:07:32 2012 +0200
Ship tests/gi in tarball
Spotted by distcheck.
tests/Makefile.am | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
commit dc2c6e6f60d2757462cbceef6176b0b3013904d3
Author: Martin Pitt <martinpitt@gnome.org>
Date: Mon Sep 3 10:18:45 2012 +0200
Fix separate build tree and distcheck
Fix regression from c7c95a79: We must set sys.path in runtests.py
properly and
cannot rely on setting $PYTHONPATH from Makefile.am only. Python
always
prepends the directory of the source file to sys.path, but that
points to the
source dir, not the build dir. The build dir has to take precedence,
otherwise
we fail to import the built libraries.
tests/runtests-windows.py | 10 +++++++---
tests/runtests.py | 13 ++++++++-----
2 files changed, 15 insertions(+), 8 deletions(-)
commit 2d8f48f4ff56bb75985136452b50b75895258608
Author: Martin Pitt <martinpitt@gnome.org>
Date: Mon Sep 3 07:57:01 2012 +0200
Split test_overrides.py
Split the huge test_overrides.py into separate files for gdk, gtk,
gio, glib,
and pango. Further split the monolithic classes for Gtk and Gio
into several
ones.
https://bugzilla.gnome.org/show_bug.cgi?id=683188
tests/Makefile.am | 5 +
tests/test_overrides.py | 2198
+----------------------------------------
tests/test_overrides_gdk.py | 119 +++
tests/test_overrides_gio.py | 114 +++
tests/test_overrides_glib.py | 445 +++++++++
tests/test_overrides_gtk.py | 1517 ++++++++++++++++++++++++++++
tests/test_overrides_pango.py | 32 +
7 files changed, 2234 insertions(+), 2196 deletions(-)
commit 1223358e2c558dd7ac3300126f989054ec5a5b3f
Author: Martin Pitt <martinpitt@gnome.org>
Date: Mon Sep 3 07:17:57 2012 +0200
_pygi_argument_to_object(): Clean up array unmarshalling
The NULL case is already handled at the top, so it does not need to be
re-checked again.
Emit a critical if we fail to allocate a Python array of the
requested size.
gi/pygi-argument.c | 47 ++++++++++++++++-------------------------------
1 file changed, 16 insertions(+), 31 deletions(-)
commit 65bfbc624bc9da6e18ff2945b14099ab8eeb7601
Author: Alban Browaeys <prahal@yahoo.com>
Date: Wed Aug 29 21:24:17 2012 +0200
Fix memory leak in _pygi_argument_to_object()
Avoid leaking the item_type_info when breaking out of the
switch in _pygi_argument_to_object() for unmarshalling arrays.
https://bugzilla.gnome.org/show_bug.cgi?id=682979
gi/pygi-argument.c | 4 ++++
1 file changed, 4 insertions(+)
commit 21b1d17d2ada2edf4063a4262b3436c279da3dc2
Author: Simon Feltman <s.feltman@gmail.com>
Date: Sat Sep 1 03:40:31 2012 -0700
Fix setting pointer fields/arguments to NULL using None.
Setting gi pointers will set them to the address of the python object.
This is good except in the case of None which should be used to NULL
the pointer out as a special case.
https://bugzilla.gnome.org/show_bug.cgi?id=683150
gi/pygi-argument.c | 8 +++++++-
tests/test_overrides.py | 41 +++++++++++++++++++++++++++++++++++++++++
2 files changed, 48 insertions(+), 1 deletion(-)
commit 6123e6f5001ca5eaea18123d8a53525abab31a45
Author: Martin Pitt <martinpitt@gnome.org>
Date: Thu Aug 23 06:44:27 2012 +0200
Fix for python 2.6, drop support for < 2.6
Replace sys.version_info.major access to tuple access which also
works for
Python 2.6.
When building for Python 2.6, inject some missing unittest API such as
@unittest.skipUnless and assertGreaterEqual() into the unittest
module in
runtests.py, so that the tests have a chance to run.
As building with Python 2.5 has been broken for a long time with
nobody
complaining, and 2.5 is ancient, bump minimum Python requirement to
2.6. Drop
obsolete #ifdef paths which only apply to <= 2.5.
https://bugzilla.gnome.org/show_bug.cgi?id=682422
configure.ac | 2 +-
gi/_glib/pyglib-python-compat.h | 8 --------
gi/_gobject/gobjectmodule.c | 8 --------
gi/module.py | 2 +-
tests/runtests.py | 27 +++++++++++++++++++++++++++
tests/test_gi.py | 2 +-
6 files changed, 30 insertions(+), 19 deletions(-)
commit b1a9848a7a7255e6b1ccd98712dd62b1514078b9
Author: Thibault Saunier <thibault.saunier@collabora.com>
Date: Tue Aug 21 07:54:09 2012 +0200
Allow overrides in other directories than gi itself
Use pkgutil.extend_path() for the gi and gi.overrides modules, so that
libraries can install overrides in a path that is different from
the one that
pygobject installs itself into. These overrides need to put this
into their
__init__.py at the top:
from pkgutil import extend_path
__path__ = extend_path(__path__, __name__)
and put themselves somewhere into the default PYTHONPATH.
https://bugzilla.gnome.org/show_bug.cgi?id=680913
Co-Authored-By: Martin Pitt <martinpitt@gnome.org>
Co-Authored-By: Simon Feltman <s.feltman@gmail.com>
gi/__init__.py | 4 ++++
gi/overrides/__init__.py | 4 ++++
tests/gi/__init__.py | 2 ++
tests/gi/overrides/Regress.py | 26 ++++++++++++++++++++++++++
tests/gi/overrides/__init__.py | 2 ++
tests/test_overrides.py | 6 ++++++
6 files changed, 44 insertions(+)
commit c7c95a795eee499373499ea5b771447746317bfb
Author: Simon Feltman <s.feltman@gmail.com>
Date: Thu Aug 23 06:03:09 2012 +0200
Clean up sys.path handling in tests
Only set sys.path once in runtests.py, not in the individual test
modules. This
reduces hidden dependencies between tests by building up a run
order dependent
search path, and also makes it easier in the future to run the
tests against
the installed system libraries.
Side issue in https://bugzilla.gnome.org/show_bug.cgi?id=680913
tests/runtests-windows.py | 6 +++++-
tests/runtests.py | 3 +++
tests/test_everything.py | 1 -
tests/test_gdbus.py | 3 ---
tests/test_overrides.py | 3 ---
tests/test_pygtkcompat.py | 3 ---
6 files changed, 8 insertions(+), 11 deletions(-)
commit 3e3525e93d852cde0f63e835b774a9b004773c69
Author: Simon Feltman <s.feltman@gmail.com>
Date: Sun Aug 19 02:30:39 2012 -0700
Fix dynamic creation of enum and flag gi types for Python 3.3
Importing Gtk was crashing on instantiation of dynamic Enum and Flag
subclasses due to what looks to be an unsupported technique. Change
tp_new() method for classes dynamically derived from PyGEnum_Type and
PyGFlags_Type to call PyLong_Type.tp_new() instead of attempting
to call
__new__() as a python method. This technique seems to work with all
versions of python so the previous python version checking also became
unnecessary.
https://bugzilla.gnome.org/show_bug.cgi?id=682323
gi/_gobject/pygenum.c | 29 ++++++++++++++++-------------
gi/_gobject/pygflags.c | 21 +++++++++------------
2 files changed, 25 insertions(+), 25 deletions(-)
commit dd31b67e821f92b5f1c2ee0382cac5edd477cd11
Author: Paolo Borelli <pborelli@gnome.org>
Date: Wed Aug 22 10:45:39 2012 +0200
[API add] Override g_menu_item_set_attribute
This C utility API take a vararg, add a corresponding override that
takes a list of tuples
https://bugzilla.gnome.org/show_bug.cgi?id=682436
gi/overrides/Gio.py | 10 ++++++++++
tests/test_overrides.py | 11 +++++++++++
2 files changed, 21 insertions(+)
commit 836902801373e386d370c44e7487aac3432f19f6
Author: Martin Pitt <martinpitt@gnome.org>
Date: Mon Aug 20 23:37:40 2012 +0200
post-release bump to 3.3.91
configure.ac | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)