| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
| |
Since we now target Python 3.8+, looking at Debian and Ubuntu releases
we have 1/2.64 in Ubuntu 20.04 and 1/2.66 in Debian bullseye.
Go with the older one of the two.
|
|
|
|
|
| |
We now require 1.60.0, to unconditionally use the GI_CHECK_VERSION
version check macro.
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
When the interface is being registered by PyGObject and again through
an override, the first one is being leaks, free it at this point.
We need to copy the GInterfaceInfo to set on the GType QData as we
do not own it.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Up until now pycairo provided a cairo.get_include() helper which
could be used to find the required include directory matching the module,
considering various scenarios.
Starting with 3.8 this leads to problems on Windows since CPython on Windows
will no longer use PATH for the DLL lookup and expects the library user to
explicitely pass the directory where the cairo DLL can be found.
In a build environment the user has no control over this though, so we have to
find the include directory without loading/importing pycairo again.
This now uses a combination of importlib.util.find_spec() for finding the module
and importlib.metadata.distribution() for finding the package version.
Hopefully this covers all cases.
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
It results in crashes downstream (pulls in a meson fallback in gst-ci)
and it isn't really needed, so revert for now.
Also Fedora only has 3.1 it turns out.
|
|
|
|
|
|
| |
Motivated by the EOL of Python 3.5 and the EOL of Ubuntu 16.04 next year
this requires Python 3.6 and moves all other dependencies to what is available in
Ubuntu 18.04.
|
| |
|
| |
|
|
|
|
| |
Depend on setuptools to make sure we write out requires-python in all cases
|
| |
|
| |
|
|
|
|
| |
Fixes #363
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
It gets used by pip>=9 to decide which version from pypi to install. Which means we can
more easily drop older Python versions without having to worry about breaking anyones setup.
Sadly our oldest supported system, Ubuntu Xenial, only has pip 8, so it will install
the newest version and fail anyway if we drop Python 3.5 support. We can at least
point users to update their pip in their virtual environments then.
https://packaging.python.org/guides/distributing-packages-using-setuptools/#python-requires
|
|
|
|
|
|
|
|
| |
The headers of Python 3.8 trigger the warning and PEP7 states that
Python is depending on this now.
As far as I remember this was mostly added to keep the code compatible
with ancient py2 MSVC, so only use it with Python 2.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
It's the one shipped with Ubuntu 16.04, which is the oldest distro
we run tests on.
|
| |
|
|
|
|
|
|
|
|
|
| |
PYGOBJECT_WITHOUT_PYCAIRO=1 python3 setup.py install
or with a recent pip:
PYGOBJECT_WITHOUT_PYCAIRO=1 pip3 install --no-build-isolation pygobject
I don't see a nice way for telling pip not to require pycairo, so this will
have to do for now.
|
|
|
|
|
|
| |
This enables the test command to be also usable on Visual Studio builds,
where we allow the test DLLs and .gir/.typelib's and test Python modules
to be built on Visual Studio builds as well.
|
|
|
|
|
|
|
|
|
|
|
| |
The distutils C Compiler object for MSVC does not have a compiler
sub-attribute, so check the compiler type before we try to use that
sub-attribute. This will fix builds on MSVC using distutils.
Also, in place of checking the various GCC/CLang compiler flags for
compile-type warnings we want to look out for, force-include
msvc_recommended_pragmas.h when we are building with Visual Studio, as
we are doing in the other GNOME projects, to achive similar effects.
|
|
|
|
| |
This reverts commit f7503c4cd1c03fde215024e61db9e1a439f39997.
|
|
|
|
|
|
| |
The compiler.compiler attribute isn't available with msvc and errors
out. The warning flag detection code only works with gcc style compilers
anyway so just skip it with msvc.
|
|
|
|
|
|
|
| |
so that _gi.so builds in a reproducible way
in spite of indeterministic filesystem readdir order
See https://reproducible-builds.org/ for why this is good.
|
|
|
|
|
|
| |
There is no non-hacky way to make this configurable for users so
this just makes it work depending on a global variable. At least
this makes it easy to patch.
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
It contains more strict checks which might be useful for finding potential bugs.
Initialize PyGILState_STATE because gcc can't figure out that state is always
defined in the release case when Py_DEBUG is enabled..
Remove -Winline which is triggerd for pygobject_init() with the debug build.
Not sure what to do when inlining fails, so just remove the warning for now.
|
| |
|
|
|
|
|
| |
We have to assign PyCFunctionWithKeywords functions to PyCFunction fields
everywhere, so not much we can do. Ideas welcome.
|
|
|
|
| |
reduces the output at build time a bit
|
|
|
|
| |
Add all the meson related files to the manifest
|