[cairo] [PATCH 2/4] configure.ac: Explicitly enable subject-objects for automake 1.11

[Bryce W. Harrington]

On Thu, Jul 10, 2014 at 09:42:15AM +0200, Uli Schlachter wrote:
> Hi,
> 
> On 10.07.2014 09:02, Bryce W. Harrington wrote:
> [..]
> >(I'm noticing a lot of misc. new warnings building with gcc 4.8,
> >particularly with --enable-qt as g++ seems fussier about things.  But
> >I've only given attention to a couple of the really noisy warnings so
> >far).
> 
> Thanks for looking into this. Debian testing recently gave me a new
> GCC version and I've just been ignoring the fact that I can't build
> cairo any more. If only the day had more hours...

Thanks for the quick reviews.  I've pushed the three patches.

> From
> 
> [...]
> >   https://bugs.freedesktop.org/show_bug.cgi?id=69874
> 
> "Yeah.  And subdir-objects is broken for the way we have our build
> set up at the moment (libdricore built from a separate directory, on
> the same .c files as core)."
> 
> That's exactly what cairo is doing with cairo-time.c. Eric Anholt
> didn't say in his comment what exactly can go wrong, but I still
> think that we need to "hack this up differently" before we can go to
> subdir-objs.
> 
> [...]
> >So yeah, looks like you're right.  I'll withdraw this patch.
> [...]
> 
> Any ideas how we could fix things properly?
> We could move all of cairo-time.c into cairo-time.h, but that seems like a hack.
> We could create a libtool convenience library containing
> cairo-time.c and link it into both places that need it, but that
> seems like too much work for too little gain.
> We could make the _cairo_time functions part of cairo's API, but
> obviously that's a bad idea.

Looks like mesa addressed it using symlinks.  Look at some of Emil
Velikov's changes to mesa late last year, for example ad501a53,
2c1bb792, etc.  Also I gather they shifted some of the shared code into
common areas.

Doing an out of tree build and then comparing the tree contents before
and after should be sufficient to spot discrepancies.

Bryce