| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fixes comiler warnings on 64bit Windows.
Python 3 now uses Py_hash_t as return value for hashfunc. Use
a compat define so we can use the right type depending on the
Python version.
On Python 2 + 64bit Windows the pointer doesn't fit the long
return type. Add a macro which handles the casting for that case
to prevent compiler warnings.
https://bugzilla.gnome.org/show_bug.cgi?id=780591
|
|
|
|
|
|
|
| |
GPid on Windows is a pointer and not int, and pointers don't fit long
on 64bit so use PyLong_AsVoidPtr/PyLong_FromVoidPtr there instead.
https://bugzilla.gnome.org/show_bug.cgi?id=780591
|
|
|
|
|
|
|
|
| |
This also adds some tests for setting flags and enum properties
which depend on some properties being added in g-i's gimarshallingtests:
https://bugzilla.gnome.org/show_bug.cgi?id=780652
https://bugzilla.gnome.org/show_bug.cgi?id=726484
|
|
|
|
|
|
| |
Gio.VolumeMonitor.get() should be used instead
https://bugzilla.gnome.org/show_bug.cgi?id=744690
|
|
|
|
|
|
| |
Move the code into override()
https://bugzilla.gnome.org/show_bug.cgi?id=686835
|
|
|
|
|
|
|
|
|
|
| |
* Instead of waiting for a fixed time, use a timeout and stop the main
loop right after the test has succeeded.
* Replace time.sleep to sync processes with os.pipe communication
* Chain idle sources instead of using multiple timeout sources
* Replace sleeps with unbufferd communication
https://bugzilla.gnome.org/show_bug.cgi?id=698548
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The code containing the bug no longer exists and the test doesn't do
what it says. The thread never gets terminated and it doesn't
assert anything. And I'm not sure what it should assert.
For the corresponding bug see
https://bugzilla.gnome.org/show_bug.cgi?id=663068
While the test doesn't hurt, it adds half a second wait time
to the test suite.
https://bugzilla.gnome.org/show_bug.cgi?id=698548
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
(in a msys2 environment)
* Replace LD_LIBRARY_PATH with gir code in runtests.py
* Remove unneeded runtests-windows.py
* Unset MSYSTEM to disable path separator hacks by msys2
* Set sys.path in runtests.py
* Skip various tests failing/hanging/crashing on Windows
https://bugzilla.gnome.org/show_bug.cgi?id=780396
|
|
|
|
|
|
|
| |
* Skip all tests which fail, crash or hang.
* Disable D-Bus in case dbus-run-session is not available.
https://bugzilla.gnome.org/show_bug.cgi?id=780396
|
|
|
|
|
|
|
|
| |
Due to the switch to AX_COMPILER_FLAGS which adds some more
warning flags. I didn't notice these as they only get triggered
on 32bit builds. Tested with gcc 6.3 and clang 3.9.
https://bugzilla.gnome.org/show_bug.cgi?id=780409
|
| |
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=773803
|
|
|
|
|
|
| |
The text was not inserted when no tags were provided.
https://bugzilla.gnome.org/show_bug.cgi?id=772896
|
|
|
|
|
|
|
|
|
|
| |
Use g_callable_info_get_instance_ownership_transfer() instead
of hardcoding the transfer for instances.
get_instance_ownership_transfer() is available since 1.42.0, so no
version dependency bump is required.
https://bugzilla.gnome.org/show_bug.cgi?id=735076
|
|
|
|
|
|
| |
Use PyGILState_Ensure/PyGILState_Release instead.
https://bugzilla.gnome.org/show_bug.cgi?id=699440
|
|
|
|
|
|
|
|
|
|
| |
This removes the --disable-thread configure option.
* glib removed disable-threads in 2011
* We use PyEval_InitThreads() for some time now which means we
build against a Python with threads. And nobody complained.
https://bugzilla.gnome.org/show_bug.cgi?id=699440
|
|
|
|
|
|
|
|
|
| |
version matches
enable_gtk() isn't idempotent and was breaking the API when called multiple times.
This ignores the call in case the version passed is the same as for previous calls.
https://bugzilla.gnome.org/show_bug.cgi?id=759009
|
|
|
|
|
|
|
|
|
|
| |
Pull LDFLAGS into variables for command and Python extension specific.
Use -shext .so for the Python extension on both Linux and OS X. The .so is
is needed by Python on OS X (not dylib).
Remove "all" target and explicit .la targets as these are handled by automake
and the "check" target on demand.
https://bugzilla.gnome.org/show_bug.cgi?id=762176
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=772949
|
|
|
|
|
|
| |
Co-Authored-By: Mikhail Fludkov <misha@pexip.com>
https://bugzilla.gnome.org/show_bug.cgi?id=772949
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=772949
|
|
|
|
|
|
| |
Use g-ir utils found by pkg-config to not fail if gobject-introspection is not installed systemwide
https://bugzilla.gnome.org/show_bug.cgi?id=772949
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Based on the patch at
https://bugzilla.gnome.org/show_bug.cgi?id=667959#c7
* Rebased on master
* Fixes the cairo_region_from_arg signature
* Add check for pycairo region support (only in pycairo master)
* Add some tests
Co-Authored-By: Martin Pitt <martin.pitt@ubuntu.com>
Co-Authored-By: Christoph Reiter <creiter@src.gnome.org>
https://bugzilla.gnome.org/show_bug.cgi?id=667959
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
aclocal needs to know where the local macros are for including them and
for having a place to put the required system ones in.
Since 1.13 it can parse configure.ac and get the path from AC_CONFIG_MACRO_DIR.
This didn't work because for some reasons it complains if the local macros
start with "AM_" even if they are there (I guess it assumes they can't be local).
To work around that, change the AM prefix to PYG, bump the automake version so
we can asume aclocal finds it and remove the hardcoded paths from both
autogen.sh and Makefile.am.
https://bugzilla.gnome.org/show_bug.cgi?id=777713
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
See https://wiki.gnome.org/Projects/GnomeCommon/Migration
Visible changes:
* --enable-compile-warnings is now [no/auto/yes] instead of
[no/minimum/yes/maximum/error]
* warnings are errors by default except for releases or if
--disable-Werror is passed (which is the default in jhbuild)
AX_APPEND_COMPILE_FLAGS() gets used to disable some warnings enabled by
AX_COMPILER_FLAGS() because they trigger with pycairo/python headers.
Tested with gcc 6.3, clang 3.8.1
https://bugzilla.gnome.org/show_bug.cgi?id=777713
|
|
|
|
|
|
|
|
| |
This is useful for a the next commit which switches away from
gnome-common and uses AX_COMPILER_FLAGS adding some new compiler
warning flags.
https://bugzilla.gnome.org/show_bug.cgi?id=777713
|
| |
|
|
|
|
|
|
|
| |
eggs are deprecated and the command was referencing long
gone files and failing, so nobody was using it.
https://bugzilla.gnome.org/show_bug.cgi?id=777719
|
|
|
|
|
|
|
|
|
|
|
|
| |
They aren't shipped, can't be build and describe the old
pygobject 2 interface.
There also was a doc-dist make target, but it was broken.
If someone wants to look things up they are in git or still
online for 2.28: https://developer.gnome.org/pygobject/stable/
https://bugzilla.gnome.org/show_bug.cgi?id=777719
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
Override Gdk.Color and Gdk.RGBA according version.
Use properly Gtk.init_check in Gtk+-4.0
https://bugzilla.gnome.org/show_bug.cgi?id=777680
|
|
|
|
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=760056
added some code triggering "missing-prototype" which we by default
treat as an error. This disables that specific error by default.
https://bugzilla.gnome.org/show_bug.cgi?id=777534
|
|
|
|
|
|
|
|
|
| |
Two new errors popped up:
* E305 expected 2 blank lines after class or function definition
* E741 ambiguous variable name
https://bugzilla.gnome.org/show_bug.cgi?id=776009
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The current behaviour is problematic because new versions of pep8/pyflakes
add new warnings or get more picky. This makes testing
hard in case newly added code is developed on an older version
or when testing old branches/releases.
This moves the pep8/pyflakes tests to a new "make check.quality"
command.
This also removes the SKIP_PEP8 flag which was added in 25d12afd06863ce
to speed up make check.
https://bugzilla.gnome.org/show_bug.cgi?id=764087
|
|
|
|
|
|
|
| |
Skip things which are gone since 4.0.
This only fixes importing and doesn't add any tests for now.
https://bugzilla.gnome.org/show_bug.cgi?id=773315
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Calling unref will cause gi and gi.repository.GLib to be imported.
However, if the program is exiting, then these modules have likely been
removed from sys.modules and will raise an exception. Assume that's the
case for ImportError and ignore the exception since everything will be
cleaned up, anyways.
This can be triggered with the following trivial program:
$ python3 -c 'from gi.repository import GLib; v = GLib.Variant("s", "foo")'
Exception ignored in:
Adding some debug code to show the full exception revealed this:
Traceback (most recent call last):
File "/home/dan/src/pygobject/build3/gi/overrides/GLib.py", line 265, in __del__
self.unref()
ImportError: import of 'gi.repository.GLib' halted; None in sys.modules
https://bugzilla.gnome.org/show_bug.cgi?id=776092
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In python3, it seems that the callable cache deinit can be called
multiple times when the program is exiting. Make that safer by clearing
the various pointers in the structure using g_clear_pointer and
Py_CLEAR. A subsequent call will skip all the deinit by seeing NULL
pointers for the members.
This was causing a critical warning when destroying the arg name hash
table multiple times with the following trivial program:
$ python3 -c 'from gi.repository import GLib; v = GLib.Variant("s", "foo")'
(process:32378): GLib-CRITICAL **: g_hash_table_destroy: assertion 'hash_table != NULL' failed
(process:32378): GLib-CRITICAL **: g_hash_table_destroy: assertion 'hash_table != NULL' failed
https://bugzilla.gnome.org/show_bug.cgi?id=776092
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Make _pygi_marshal_from_py_filename handle None input
values. This allows one to pass None to parameters
annotated as nullable filenames.
This fixes a test suite error in test_spawn_async_with_pipes
triggered by an annotation change in glib.
https://bugzilla.gnome.org/show_bug.cgi?id=770821
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
glist and ghashtable objects both store pointers. Complex objects are
stored as pointers to the objects, but simpler objects like an integer
value are stored directly as a pointer, using for example the
GINT_TO_POINTER and GPOINTER_TO_INT macros.
This is done in pygobject with the _pygi_hash_pointer_to_arg and
_pygi_arg_to_hash_pointer functions. These functions handle the various
type of objects. However they consider that an enum, represented with the
GI_TYPE_TAG_INTERFACE type (extended interface object), are always a
pointer. This is wrong as it is often a 32-bit value. Therefore on 64-bit
big endian machines, the value is handle with the 2 32-bit parts swapped.
This patches fixes that by changing the second argument of both functions
from GITypeTag to GITypeInfo. This way the interface can be determined,
and the underlying storage type can also be determined. This currently
only handles enum and flags, leaving other types as pointers. The patch
also adds two tests in the testsuite, one for each direction.
https://bugzilla.gnome.org/show_bug.cgi?id=770608
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=766396
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The use of dbus-launch is deprecated by dbus upstream.
See https://lists.debian.org/debian-devel/2016/08/msg00554.html for more
details.
So use dbus-run-session instead, which was specifically designed to be
used for such cases, like running test-suites.
See also https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=836049.
https://bugzilla.gnome.org/show_bug.cgi?id=770798
|
| |
|