Since all repo URLs and tags are following a common scheme, we can
reduce boilderplate by computing them by driver and and version only.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
> In file included from ../glx/glxdricommon.c:35:
> /usr/local/include/GL/internal/dri_interface.h:445:26: warning: redefinition of typedef 'GLsync' is a C11 feature [-Wtypedef-redefinition]
> 445 | typedef struct __GLsync *GLsync;
> | ^
> /usr/local/include/GL/glext.h:1367:26: note: previous definition is here
> 1367 | typedef struct __GLsync *GLsync;
> | ^
Since it's coming from MESA, the only thing we can do here is
suppressing this warning explicitly.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
We're not using gitlab anymore (and the CI config there is totally
unmaintained), so better move the scripts to the right place.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Added a workflow to check the commit messages of a PR to make sure each is
signed off by the author. A repo token is required, here named SECRET_TOKEN,
which has access to read and write PR comments.
Signed-off-by: JSOwens <josephs.owens@gmail.com>
f.d.o is failing too often in recent times, so switching to our
own mirrors at github.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Their latest releases have still configured wrong submodule URLs,
so we need to use master here.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
MacOS build suddenly missing xvfb-run, which did work up until few hours ago.
Therefore, make a quick workaround, until the situation is clearly resolved.
The only impact is that some XTS tests (which are ignored on MacOS anyways)
might not work properly.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Make sure the release message won't be created when build jobs on the
release tag didn't succeed.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Now that the test suite has been fixed to work on the CI, we no longer
need (and should) ignore broken test runs.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
The GLES tests need an actual GPU (/dev/dri/* device), which is not available
within github CI runners, so we need to skip those when running there.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
We need to run the Xts build inside Xvfb, because it needs a running
Xserver for creating tetexec.cfg.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Also store the piglit results into the build artifacts, so we can now
easily look at them later.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Several sites we need to clone our dependencies from (eg. freedesktop.org)
have pretty unreliable servers, so our CI jobs often fail just because
of temporary clone failure.
Therefore adding a separate cloning script, which is more clever with automatic
retries, but it also tries to keep the traffic low (eg. trying shallow clones
if possible) and automatically detecting whether we're pulling a ref or a
direct commit.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
The mingw32 build job often fails due frequent temporary problems on
cloning vom savannah. Retry several times, hoping to fix it.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
