There's no need for high performance in these little tests, so instead of
supporting various platform-specific sendfile() implementations, just use a local read-write function.
Doing so means we can ditch the specific input list for doxygen and just copy
all files over into our builddir/doc/ subdir, then use that subdirectory as
input data.
This relies on meson putting a subdir() into a subdirectory in the build
directory. This isn't technically guaranteed but I also suspect that if meson
ever changes that, lots of other projects will break. Even in that case we
should build just fine since we now filter for *.h and *.dox and don't copy
any other doxygen-commented files into the builddir anyway.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This hasn't been a real .in file since the autotools removal, so rename it to
reflect that. And since we can call it with arguments from meson, let's do so
in the most sensible manner - passing the full paths in as required rather
than relying on a directory layout within the script.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
The only trackpoint where I have a rough idea what we need is the ALPS v8 one.
All other ones we'll have to re-do.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Measuring the trackpoint range has not shown to be sufficient or precise
enough to be used as an ingredient for trackpoint acceleration. So let's just
switch back to a generic multiplier that we can apply to the input deltas do
undo any device-specific lack of scaling.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This gets rid of the trackpoint range propery that we've been parsing until
now and instead just opts for a basic curve with some deceleration for low
pressure. The speed range is taken from the touchpad and should be wide enough
for most trackpoints that fall within the expected range.
Trackpoints like the new ALPS ones need to be configured through a hwdb (this
part is currently missing).
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
When the meson build type is something other than the debug types, we don't
need the special behavior where we adjust executable paths and data dir
lookup for tools run directly from the builddir.
This avoids leaking the build dir into the final executables.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Doesn't actually do anything but this way they end up in the builddir and can
be picked up by ./builddir/libinput measure fuzz, etc.
And rename the source files to .py to signal that they are not supposed to be
directly executed.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Don't use a custom hack here, just make sure the tool ends up in the builddir
so it's picked up by the libinput main tool.
This means the PATH isn't set up correctly when called directly
(./builddir/libinput-measure-touchpad-pressure) but the workaround is to
always use the libinput tool - just as we expect from users.
To make it more obvious that we're not supposed to run this directly, rename
the source file to .py
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
When running libinput tools from the builddir, look up the subtools in the
builddir as well. Otherwise, add the install prefix to the list of lookup
locations.
This ensures that a) we're running builddir stuff against builddir stuff, but
also b) that we're not running builddir stuff against installed stuff because
that may give us false positives.
The test was squashed in from a separate patch and was
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
The libinput/doc/latest is now built automatically from git, so it's good to
have a reference to show which commit it was being built from. Add a section
to the readme with git commit information that is replaced by meson.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Using files() over a list of hand-constructed paths is the recommended way.
But unfortunately doxygen needs its input files as a string list, so we still
have to build that list anyway. Still, this way we don't need to hardcode
every file with the source root, we just assemble it as we go.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This is by far the slowest job, move it up so it gets started earlier and we
have more parallelization going on.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
I keep hitting this in the gitlab runner, 100ms is clearly not enough here for
slow containers.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Slightly more obvious than just "html". Main motivation here is that we want
to provide the documentation as artifact from GitLab's CI, so having it unzip
to something slightly more meaningful makes sense.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
We don't run it on all machines, valgrind fails right now on arch because of a
memleak in bash itself. To avoid having CI failures that aren't our fault,
only run on F28 because that's the one I'm tracking.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
