Re: [XEN PATCH 2/2] automation: Check if ninja is available before building QEMU

[Anthony PERARD <anthony.perard@xxxxxxxxxx>]

On Wed, Jul 07, 2021 at 06:46:03PM +0100, Andrew Cooper wrote:
> On 07/07/2021 17:40, Anthony PERARD wrote:
> > ninja is now required to build the latest version of QEMU, some
> > container still don't have ninja and attempting to add it breaks the
> > build for different reasons, so QEMU will be skip on those containers.
> >
> > Failures:
> > - ubuntu/xenial:
> >     fatal: ninja version (1.5.1) incompatible with build file 
> > ninja_required_version version (1.7.1).
> > - debian/unstable-i386: (debian-unstable-32)
> >     /build/tools/flask/policy/Makefile.common:115: *** target pattern 
> > contains no '%'.  Stop.
> > - suse/opensuse-tumbleweed:
> >     failed to build iPXE
> > - debian/unstable:
> >     update of the container fails with:
> >         The following packages have unmet dependencies:
> >          clang-8 : Depends: libstdc++-8-dev but it is not installable
> >                    Depends: libgcc-8-dev but it is not installable
> >                    Depends: libobjc-8-dev but it is not installable
> >                    Recommends: llvm-8-dev but it is not going to be 
> > installed
> >                    Recommends: libomp-8-dev but it is not going to be 
> > installed
> 
> Only the first failure seems to be related to ninja.
> 
> The second is a known bug which still needs addressing.  Our build
> system's logic to cope with `checkpolicy` being absent doesn't work.
> 
> Tumbleweed is generally broken and fails at ./configure due to missing
> compression libraries.  It is quite possible that the iPXE failure is
> related.
> 
> The final one is probably because unstable has dropped clang-8.  The
> container hasn't been rebuilt in a while because we still haven't got
> auto-rebuild sorted.


So those containers are ones that I've noticed needed ninja (as found
out by looking at a single pipeline), but one which trying to install
ninja and thus rebuilding the container mean that the build was broken.
I just didn't want to spend time trying to fix issues that should have
been found out by a regular rebuild of the containers. So I've just
listed the reason there and moved on.

-- 
Anthony PERARD