[libdrm PATCH 2/2] configure.ac: don't detect disabled options dependencies

[Marcin Ślusarz]

On Tue, Jan 19, 2016 at 10:26:16AM +0200, Emil Velikov wrote:
> Hi Marcin,
> 
> On 9 January 2016 at 17:05, Marcin Ślusarz <marcin.slusarz at gmail.com> wrote:
> > Currently with --disable-amdgpu --disable-valgrind --disable-cairo-tests
> > cunit, valgrind and cairo are still detected.
> >
> Doesn't this patch make 1/2 obsolete ?

Yes, more or less. Both patches fix the issue I was seeing.

> 
> > Signed-off-by: Marcin Ślusarz <marcin.slusarz at gmail.com>
> 
> Reviewed-by: Emil Velikov <emil.l.velikov at gmail.com>
> 
> > ---
> >  configure.ac | 36 ++++++++++++++++++++++--------------
> >  1 file changed, 22 insertions(+), 14 deletions(-)
> >
> > diff --git a/configure.ac b/configure.ac
> > index c8c4ace..cc44e3f 100644
> > --- a/configure.ac
> > +++ b/configure.ac
> 
> > @@ -400,7 +404,9 @@ AC_ARG_ENABLE([cairo-tests],
> >                [AS_HELP_STRING([--enable-cairo-tests],
> >                                [Enable support for Cairo rendering in tests (default: auto)])],
> >                [CAIRO=$enableval], [CAIRO=auto])
> > -PKG_CHECK_MODULES(CAIRO, cairo, [HAVE_CAIRO=yes], [HAVE_CAIRO=no])
> > +if test "x$CAIRO" != xno; then
> > +       PKG_CHECK_MODULES(CAIRO, cairo, [HAVE_CAIRO=yes], [HAVE_CAIRO=no])
> > +fi
> Namely: if we have disable-cairo-tests, the module will never be
> checked, thus CAIRO_{CFLAGS,LIBS} will be empty. Obviously the user
> can explicitly sets the CAIRO_* variables, in which case they're
> cutting the branch they're standing on.
> 
> Please let me know if you'd prefer to have 1/2 regardless (or if I
> missed something)

I thought it would be better to have cairo build flags guarded by global
HAVE_CAIRO flag. If you want to apply just this patch, that's fine by me.

> 
> Thank you for sorting these.

Thanks,
Marcin