Re: [OE-core] [PATCH 0/4] pulseaudio: Autotools to Meson

["Tanu Kaskinen" <tanuk@iki.fi>]

On Sat, 2020-12-19 at 12:36 -0800, Khem Raj wrote:
> On Sat, Dec 19, 2020 at 9:24 AM Khem Raj <raj.khem@gmail.com> wrote:
> > On Thu, Dec 17, 2020 at 11:28 AM Tanu Kaskinen <tanuk@iki.fi> wrote:
> > > The PulseAudio project plans to drop Autotools support in 15.0, so let's
> > > switch to Meson. The first patch removes OE_LT_RPATH_ALLOW, which is
> > > some old cruft that sounds like it has something to do Libtool, so it's
> > > perhaps relevant for the goal of dropping Autotools. The next two
> > > patches disable features that aren't available with Meson. The last
> > > patch does the build system switch.
> > > 
> > > Tanu Kaskinen (4):
> > >   pulseaudio: Remove OE_LT_RPATH_ALLOW
> > >   pulseaudio: disable EsounD support
> > >   pulseaudio: disable GConf support
> > >   pulseaudio: switch build system from Autotools to Meson
> > 
> > I am seeing build failures with this patch series while using clang
> > for compiler see.
> > 
> > https://errors.yoctoproject.org/Errors/Details/539474/
> > 
> 
> It turns out that new meson checks are a bit different and if
> -Qunused-arguments is used with clang then the test for neon passes
> because its a simple compile time check in meson, So a workaround that
> I have done is to remove using -Qunused-arguments on clang
> cmdline and then it works ok.

Ok, so the Neon check in Meson is somehow too lax if it passes with
-Qunused-arguments. Thanks for the initial debugging! I can look into
the issue, but I can't promise to do it quickly. I see these patches
are in master already - should the last patch be reverted for now? Or
can your workaround be applied to master (or meta-clang, if that's
where clang workarounds belong)?

-- 
Tanu