Prior to this change, the Git version was only getting generated when
the call was made to autoreconf. This was resulting in stale version
info.
This change, by contrast, generates the build info at the call to make.
While this doesn't completely eliminate the stale version issue, it's
easier to work around, by simply running `make clean`, `make`.
Prior to this change, all the generated HTML documentation and locale
files were using the `x52pro-linux` name. That name is no longer
reflective of the project, since it works on macOS and with the non-Pro
X52 as well.
This change is also reflective of the patch used in the released PPA,
which takes care of this anyway, but is being added to the sources to
better integrate with PKGBUILD/AUR/MPR.
When building from source, it is desired that we embed the version
string into the resulting binaries so that we can determine exactly what
version of the sources were used.
This change adds a Version file, which always holds the latest release
version, and a configure.version script, which tries to get the version
information from Git, before falling back to using the version embedded
in the above file. The generated configure script will then have the
version embedded within it, which will then create the BUILD_VERSION
definition in config.h. Applications can then use this definition as
needed.
Prior to this change, the user needed to install inih as a dependency,
either from the distribution repositories, or from source. On some
platforms (notably macOS), inih is not available prepackaged, and must
be installed by the user. This tends to cause needless friction.
This change imports the ini.c and ini.h files from the upstream inih
repository into the X52 source tree. This will allow us to build the
repository on any system with the original set of dependencies, and not
have to force the user to install packages themselves.
AC_CHECK_HEADERS and AC_SEARCH_LIBS do not abort when the header or
library respectively are not found. Since inih is a hard requirement,
make sure that the action-if-not-found fields call AC_MSG_ERROR.
Fixes#32
Commit 2ce9ff22 replaced the calls to AX_PKG_CHECK_MODULES with
PKG_CHECK_MODULES, but it was done blindly, since PKG_CHECK_MODULES
takes a different number of arguments. As a result, even if the checks
for cmocka and udev succeed, they are still disabled because the
corresponding have_ variable is not set to 'yes'.
This change adds the daemon configuration parser and command line
argument parser. This also adds the associated strings to the
translation files, and integrates the daemon into the existing autotools
build framework.
Prior to this change, the udev rules were fixed to allow read-write
access only to the `plugdev` group. However, while this is the default
group for USB input devices on Debian and its derivatives, this is not
true for some systems such as openSUSE, which use `input` as the group.
This change adds a `--with-input-group` argument to `configure`, which
defaults to `plugdev`. This change also updates the install
documentation to reflect the new options.
Prior to this change, libx52 tests needed a linker that supported the
--wrap argument. This is not available on OSX, and therefore, we had to
disable the unit tests on non-Linux systems.
Since we needed to rebuild the libx52 library anyway for the test, it is
simpler to just define libusb_control_transfer to point to our mock
function. This allows us to verify libx52 on all supported platforms.
Prior to this change, we needed to add a manual override function to
mock the vendor command. Given that cmocka has built-in support for
mocking functions, it's better to use that instead.
This change simply removes the manual override and any tests that rely
on it.
Prior to this change, generation of the character lookup table could use
Python 2 or Python 3, depending on what the first instance of python in
the PATH pointed to. On most systems, python is a symbolic link to
python2.7. However, given that Python 2 is EOL, it makes sense to switch
the code to use Python 3 now, rather than later.
This change updates the requirements to use Python 3.6 or later, and
updates the Makefile to invoke the script with the detected python
binary, rather than relying on the shebang to use the system Python.
This change adds additional compiler warnings as detected by the
autoconf archive. Because the code is susceptible to these additional
warnings, disable treating warnings as errors for now, until we can fix
the warnings.
This commit changes the local timezone calculation to use tm.tm_gmtoff,
and falling back to calculating it manually from the output of strftime
when tm_gmtoff is not a member of struct tm.
This also fixes the tests so that they are no longer expected to fail.
Fixes#20.
The hope was that I could compile some tests that would override
libx52_vendor_command, and run those on OSX to bypass the skipped tests
that used x52cli. However, a Travis-CI run indicated that the compiler
on OSX doesn't support weak symbols, which renders the point moot.
This change allows exporting libx52_vendor_command as a weak symbol,
thereby allowing it to be overridden by a test runner to validate that
the library is indeed behaving as per the spec.
Because this is something that may not be necessarily desirable on a
production environment, add a configure time flag to disable building
with weak symbols. This will also disable any tests that may rely on
libx52_vendor_command being a weak function.
Prior to this change, the tests were using the automake simple test
harness. The limitation was that for each set of test cases (e.g.
timezone tests), even if one test failed, it would report every test in
that set as failed. Migrating to TAP allows fine-grained reporting on
every single test case, and allows better investigation into checking
which individual tests failed.
This change also updates the timezone tests to explicitly mark the ones
that check when the timezone is Pacific Daylight Time as expected to
fail. It also updates the clock tests to improve the formatting of the
test case identifier.
nirenjan