[PATCH v1 weston 05/11] tests: Allow tests to use customized command line parameters

[Pekka Paalanen]

On Wed, 19 Nov 2014 15:06:20 -0800
Bryce Harrington <bryce at osg.samsung.com> wrote:

> From: Derek Foreman <derekf at osg.samsung.com>
> 
> Tests will now return the extra command line parameters they need
> when run with --params
> 
> Signed-off-by: Bryce Harrington <bryce at osg.samsung.com>
> ---
>  tests/weston-test-runner.c | 8 ++++++++
>  tests/weston-tests-env     | 1 +
>  2 files changed, 9 insertions(+)
> 
> diff --git a/tests/weston-test-runner.c b/tests/weston-test-runner.c
> index ef45bae..ce0a670 100644
> --- a/tests/weston-test-runner.c
> +++ b/tests/weston-test-runner.c
> @@ -34,6 +34,8 @@
>  
>  #define SKIP 77
>  
> +char __attribute__((weak)) *server_parameters="";
> +
>  extern const struct weston_test __start_test_section, __stop_test_section;
>  
>  static const struct weston_test *
> @@ -154,6 +156,12 @@ int main(int argc, char *argv[])
>  			exit(EXIT_SUCCESS);
>  		}
>  
> +		if (strcmp(testname, "--params") == 0 ||
> +		    strcmp(testname, "-p") == 0) {
> +			printf("%s", server_parameters);
> +			exit(EXIT_SUCCESS);
> +		}
> +
>  		t = find_test(argv[1]);
>  		if (t == NULL) {
>  			fprintf(stderr, "unknown test: \"%s\"\n", argv[1]);
> diff --git a/tests/weston-tests-env b/tests/weston-tests-env
> index e332354..aaf3ee1 100755
> --- a/tests/weston-tests-env
> +++ b/tests/weston-tests-env
> @@ -46,5 +46,6 @@ case $TESTNAME in
>  			--shell=$SHELL_PLUGIN \
>  			--log="$SERVERLOG" \
>  			--modules=$TEST_PLUGIN,$XWAYLAND_PLUGIN \
> +			$($abs_builddir/$TESTNAME --params) \
>  			&> "$OUTLOG"
>  esac

Hi,

this is pretty clever. I do wonder though if we are going to need a
test-specific weston.ini at some point, too. If we do, do we also need
to control command line options? Well, apart from specifying the config
file location.

While this patch is fine for now, we are probably going to need
something more flexible soon. Currently we only run tests on a single
backend at a time, and a single shell at a time. There will become time
when we need to test different shell plugins. Oh hey, but this could
also work for shell plugins, as tests would likely be shell-specific.

It might also be nice to have a bit of heuristics and by default run on
all backends that might work:
- always headless with each renderer
- if DISPLAY is set, x11-backend
- if WAYLAND_DISPLAY is set, wayland-backend
- if neither DISPLAY nor WAYLAND_DISPLAY set, drm-backend, if DRM
  available
Hm, or maybe that doesn't make much sense... but being able to list
which all backends should run through all (the specified) tests would
be nice.

Oh well, random thoughts for the future, nothing to complain about
this now. Anyway, I started to look at this series and got this far this
week, I wanted to say. Looks good so far. :-)

I figured I want to merge this series first, so we have some tests
in place when I look at the matrix rewrite series, since that has some
potential to break some rendering.


Thanks,
pq