Whenever a new gcc version is released that introduces new warnings,
this breaks lots of builds on the user's side. This change pushes the
detection of these new warnings a bit back until either a user reports
warnings or developers get the new compiler updates. I think this is a
good tradeoff since release builds are no longer totally broken as soon
as a new compiler version comes out.
Travis still uses -Werror because there we actually want builds to fail.
The polybar executable with RelWithDebInfo gets over 100MB. And there
really isn't any reason to have users install such huge executables.
Release build type gives you approx. 3.6M executables
Fixes#1497
Fixes compilation under GCC 9
The default copy constructor implicit generation is deprecated by C++ standard.
The window& operator=(const xcb_window_t win); operator seems to be useless.
Fixes#1728
Ref jaagr/xpp#16
As the kernel documentation said:
https://www.kernel.org/doc/Documentation/ABI/stable/sysfs-class-backlight
This file will also show the brightness level stored in the driver, which may not be the actual brightness (see actual_brightness).
Therefore the brightness value should be read in the actual_brightness file.
Fix#1180
rtd scans the project for a conf.py file so we cannot name it conf.py.in
unless we get rtd to run cmake before building.
The easier option is to have doc/conf.py be the file used by rtd and all
other builds use cmake to first configure it.
This also moves the doc generation completely into cmake (no more
Makefile).
To generate the docs the project needs to first be configured and then
`make doc` can be run.
The approach used is leaned on the cmake's project own use of Sphinx:
Utilities/Sphinx/CMakeLists.txt
We need to have the version string available in multiple places not just
the source code. It is now hardcoded in the root CMakeLists.txt and all
files that need it will be configured with cmake.
This also removed the unecessary duality of GIT_TAG and APP_VERSION and
GIT_TAG_NAMESPACE and APP_VERSION_NAMESPACE.
This allows us to also have the manual as part of the generated html
documentation.
The syntax is also easier to use.
Right now the man page installed on the system is not replaced with the
rst file, this will come in a later step
The issue tracker should be for bugs only. Right now actual bugs can
easily fall through the cracks because we get stuck answering questions
and basically providing tech support.
This move is also intended to have more traffic to our subreddit and IRC
channel with the hope that the community can also help itself out.
To avoid polybar from being killed by SIGUSR1 during reloading, SIGUSR1 is ignored until the signal is registered in the new polybar process.
As stated in signal(7) man page, the ignored signals are still ignored after a call to a function of the execvX family.
During an execve(2), the dispositions of handled signals are reset to the default;
the dispositions of ignored signals are left unchanged.
Fixes#428
