Wayland and Weston 1.0.3 releases out

[Pekka Paalanen]

On Wed, 19 Dec 2012 09:20:18 +0100 (CET)
Jan Engelhardt <jengelh at inai.de> wrote:

> 
> On Wednesday 2012-12-19 08:27, Pekka Paalanen wrote:
> >> make[3]: Entering directory `/home/abuild/rpmbuild/BUILD/weston-1.0.3/tests'
> >> FAIL: surface-test.la [..].
> >> ==========================================================================
> >> 6 of 6 tests failed
> >
> >could we see the logs from the tests? I mean the files that go into
> >$abs_topbuilddir/tests/logs.
> 
> Ah, there they are - that already resolves my case. Of course you can
> see them. In various stages,

Thanks, this looks very helpful, especially for other packagers!

> ---8<---
> # button-test-serverlog.txt
> Date: 2012-12-19 UTC
> [08:02:10.967] weston 1.0.3
>                http://wayland.freedesktop.org/
>                Bug reports to:
>  https://bugs.freedesktop.org/enter_bug.cgi?product=weston
>                Build:  
> [08:02:10.968] OS: Linux, 3.4.18-jng6-default, #1 SMP Wed Nov 14 17:04:47 UTC
>  2012 (21e90c1), x86_64
> [08:02:10.968] fatal: environment variable XDG_RUNTIME_DIR is not set.
> Refer to your distribution on how to get it, or
> http://www.freedesktop.org/wiki/Specifications/basedir-spec
> on how to implement it.
> --->8---
> 
> Solution: creating a dummy XDG_RUNTIME_DIR. (Invocation is
> `mkdir -m go-rwx xdg; XDG_RUNTIME_DIR=$PWD/xdg make check`.
> Next,
> 
> ---8<---
> # button-test-log.txt
> couldn't open /home/abuild/.config/weston.ini
> couldn't open /home/abuild/.config/weston.ini
> Error:    failed to add default include path /usr/share/X11/xkb
> --->8---
> 
> Solution: install xkeyboard-config (this was inside a trimmed chroot
> for distro package building). It seems to forgive me on the missing
> weston.ini. Next,
> 
> ---8<---
> # button-test-serverlog.txt
> Date: 2012-12-19 UTC
> [08:11:03.940] weston 1.0.3
>                http://wayland.freedesktop.org/
>                Bug reports to:
>  https://bugs.freedesktop.org/enter_bug.cgi?product=weston
>                Build:  
> [08:11:03.940] OS: Linux, 3.4.18-jng6-default, #1 SMP Wed Nov 14 17:04:47 UTC
>  2012 (21e90c1), x86_64
> [08:11:03.941] Loading module
>  '/home/abuild/rpmbuild/BUILD/weston-1.0.3/tests/../src/.libs/wayland-backend.so'
> [08:11:03.943] failed to create display: No such file or directory
> [08:11:03.943] fatal: failed to create compositor
> --->8---
> 
> When run without make, the command line output is
> 
> ---8<---
> ~/rpmbuild/BUILD/weston-1.0.3/tests> XDG_RUNTIME_DIR=$PWD/xdg \
> 	./button-test 
> button-test: weston-test-client-helper.c:463: client_create: Assertion
> `client->wl_display' failed.
> test "simple_button_test":      signal 6, fail.
> 1 tests, 0 pass, 1 fail
> --->8---
> 
> Would be a good idea to print the file/directory name involved -
> I assume it tries to create/connect to a socket (similar to
> that of /tmp/.X11-unix/X0). strace eventually tells me.
> 
> ---8<---
> # XDG_RUNTIME_DIR=$PWD/xdg strace ./button-test
> [...]
> [pid 17011] connect(3, {sa_family=AF_FILE,
> sun_path="/home/abuild/rpmbuild/BUILD/weston-1.0.3/tests/xdg/wayland-0"}, 63) =
> -1 ENOENT (No such file or directory)
> --->8---
> 
> So, who is to be starting the Wayland instance? I think, the
> testsuite should do that, since it cannot rely on any preexisting
> wayland running, nor should it toy around with any (preexisting)
> process that might be used productively.

The script tests/weston-tests-env should start the compositor for each
test, but above, your quote from the compositor:
> [08:11:03.941] Loading module
>  '/home/abuild/rpmbuild/BUILD/weston-1.0.3/tests/../src/.libs/wayland-backend.so'
> [08:11:03.943] failed to create display: No such file or directory
> [08:11:03.943] fatal: failed to create compositor

It is trying to use the Wayland backend for the test compositor, which
obviously fails. We have not hooked up the headless backend to the
tests yet, as there are some related bugs, so we have to use a real
backend. It looks like the check should be run under X or another
Wayland server for now.

I suspect the script may have a typo and the fallback backend should be
the DRM backend instead of the Wayland backend.

Oh, the headless backend is not even in the 1.0 stable series. It means
it will be an option starting from 1.2 of stable series.


Thanks,
pq