[libvirt PATCH 9/9] spec: Introduce arches_*

[Neal Gompa]

On Tue, Oct 6, 2020 at 6:41 AM Andrea Bolognani <abologna at redhat.com> wrote:
>
> On Mon, 2020-10-05 at 20:40 -0400, Neal Gompa wrote:
> > On Mon, Oct 5, 2020 at 2:41 PM Andrea Bolognani <abologna at redhat.com> wrote:
> > >  %if 0%{?fedora}
> > > +    %define arches_qemu_kvm     %{arches_x86} %{power64} s390x %{arm} aarch64
> > >  %else
> > > +    %define arches_qemu_kvm     x86_64 %{power64} aarch64 s390x
> > >  %endif
> >
> > This conditional is functionally irrelevant. The superset defined for
> > Fedora does not change how things work for RHEL, and it'd be easier to
> > just use the one architecture set.
>
> The difference I can see is that %{ix86} is not currently included in
> %{arches_qemu_kvm} on RHEL, but with your change it would and, unlike
> what happens for 32-bit ARM, RHEL packages are actually being built
> on i686.
>
> Later on we have
>
> > >  %define with_storage_gluster 0%{!?_without_storage_gluster:1}
> > > +%ifnarch %{arches_qemu_kvm}
> > >      # gluster is only built where qemu driver is enabled on RHEL 8
> > >      %if 0%{?rhel} >= 8
> > >          %define with_storage_gluster 0
>
> and AFAICT that would break with your proposed change, because we
> would try to build with gluster support on i686 RHEL where gluster is
> not actually available.
>

Then can we flip this conditional to %if 0%{?rhel} for the
architecture list? As it is, it's unclear that the reason that *RHEL*
is the less-capable variant.


-- 
真実はいつも一つ！/ Always, there's only one truth!