Re: [PATCH] tests/docker/dockerfiles/centos: Use SDL2 instead of SDL1

["Daniel P. Berrangé" <berrange@redhat.com>]

On Wed, Oct 21, 2020 at 01:29:10PM +0200, Thomas Huth wrote:
> On 21/10/2020 11.38, Philippe Mathieu-Daudé wrote:
> > On 10/21/20 9:23 AM, Thomas Huth wrote:
> >> We do not support SDL1 in QEMU anymore. Use SDL2 instead.
> >>
> >> Signed-off-by: Thomas Huth <thuth@redhat.com>
> >> ---
> >>   tests/docker/dockerfiles/centos7.docker | 2 +-
> >>   tests/docker/dockerfiles/centos8.docker | 2 +-
> >>   2 files changed, 2 insertions(+), 2 deletions(-)
> >>
> >> diff --git a/tests/docker/dockerfiles/centos7.docker
> >> b/tests/docker/dockerfiles/centos7.docker
> >> index 8b273725ee..6f11af1989 100644
> >> --- a/tests/docker/dockerfiles/centos7.docker
> >> +++ b/tests/docker/dockerfiles/centos7.docker
> >> @@ -31,7 +31,7 @@ ENV PACKAGES \
> >>       perl-Test-Harness \
> >>       pixman-devel \
> >>       python3 \
> >> -    SDL-devel \
> >> +    SDL2-devel \
> >>       spice-glib-devel \
> >>       spice-server-devel \
> >>       tar \
> >> diff --git a/tests/docker/dockerfiles/centos8.docker
> >> b/tests/docker/dockerfiles/centos8.docker
> >> index 585dfad9be..f86931f955 100644
> >> --- a/tests/docker/dockerfiles/centos8.docker
> >> +++ b/tests/docker/dockerfiles/centos8.docker
> >> @@ -2,7 +2,7 @@ FROM centos:8.1.1911
> >>     RUN dnf -y update
> >>   ENV PACKAGES \
> >> -    SDL-devel \
> >> +    SDL2-devel \
> >>       bzip2 \
> >>       bzip2-devel \
> >>       dbus-daemon \
> > 
> > I wonder if we are having docker image caching issues...

Looks more like something related to the artifact caching, rather
than container image caching.

> > 
> > build-system-centos job passed but then in check-system-centos:
> > 
> > FAILED: libcommon.fa.p/ui_sdl2-input.c.o
> > ui/sdl2.h:7:10: fatal error: SDL.h: No such file or directory
> >  #include
> >           ^~~~~~~
> > 
> > And then in acceptance-system-centos:
> > 
> > ./qemu-system-x86_64: error while loading shared libraries:
> > libSDL2-2.0.so.0: cannot open shared object file: No such file or directory
> 
> Uh, weird, now I've got that one, too ... sorry, I only waited for the build
> phase to complete before sending out the patch, I did not expect any trouble
> in the testing stage...
> 
> The check-system-xxx jobs are apparently recompiling lots of files now ...
> not sure yet why that happens... seems to be related to Paolo's PR from last
> weekend?

Presumably this is related to this PR removing use of make, and instead
building everything directly with ninja.

> The previous CI run still looks okayish:
> 
>  https://gitlab.com/qemu-project/qemu/-/jobs/796502985
> 
> then the one from Paolo suddenly recompiled pretty much everything:
> 
>  https://gitlab.com/qemu-project/qemu/-/jobs/796767669
> 
> Does anybody got a clue what happened here?

The log shows that it decided "meson" needed re-running, and as a result
of re-running meson, we in turn rebuild all source.

Best guess is that somewhere in here we are not correctly preserving
timestamps on something, causing  ninja to consider the build out of
date.

Possibly this is because the the check/acceptance jobs re-clome the
git checkout again ? Coud that make ninja thing the sources have
changed, or does git clone set timestamps to match the git commit
date ?

Regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|