Re: [Ltib] Cairo fails to build

[Stuart Hughes]

Hi Guillaume,

The thought behind this is that normally you would need xlib unless you
had DirectFB.  This could be an incorrect assumption on my part.  How
does this work with xlib or directfb?

BTW: you do have DirectFB available in LTIB, this is the only way I've
actually done (minimal) testing.  To be honest the whole graphics thing
is just experimental right now.

For now I think your hack is okay to get going until I understand a bit
more about how to fix this properly.  Also note that even with xlib
enabled, it should pull in pieces from the host, so I need to
investigate why that is.  It is likely that this needs inhibiting in the
package.

Regards, Stuart

On Tue, 2008-04-01 at 18:35 +0200, Guillaume Emont wrote:
> In the .spec file for cairo, there is:
> XTRA_OPTS="--disable-win32"
> if [ -n "$PKG_DIRECTFB" ]
> then
>     XTRA_OPTS="$XTRA_OPTS --enable-directfb --disable-xlib"
> fi
> ./configure --prefix=%{_prefix} --host=$CFGHOST --build=%{_build}
> $XTRA_OPTS
> 
> I don't have directFB, but I don't want xlib either. If I add
> --disable-xlib to the configure options, cairo compiles. Is there a
> $PKG_X or $PKG_XLIB that could be used like $PKG_DIRECTFB to
> enable/disable xlib? What I did is a quick hack (added --disable-xlib at
> the definition of XTRA_OPTS), and it works, and I guess that could be
> changed into a proper correction.
> 
> On my system, it seems cairo's ./configure finds some xlib stuff from
> the host system, which triggers the inclusion of /usr/include that f***s
> up everything.
> 
> Guillaume
> 
> 
> Le mardi 01 avril 2008 à 17:58 +0200, Guillaume Emont a écrit :
> > Hi,
> > I work with Phillipe, and meets the same issue when trying to compile
> > Cairo. I think we both have the same (or very similar) setup. My host is
> > a Ubuntu gutsy, and I use the BSP from freescale
> > (ltib-imx31ads-20071008) with a few .spec files taken from the cvs
> > version of LTIB (glib2, cairo, pango and fontconfig at least).
> > I have found two .la files in rootfs/usr/lib: libbfd.la and
> > libopcodes.la. Rpm says they don't belong to any package, and removing
> > them don't change the issue...
> > Yet, in the gcc command launched by make, there is a *very* suspicious
> > -I/usr/include. I will have a look at cairo's autotools magic and try to
> > see if I understand where it comes from.
> > 
> > Cheers,
> > 
> > Guillaume
> > 
> > 
> > 
> > _______________________________________________
> > LTIB home page: http://bitshrine.org
> > 
> > Ltib mailing list
> > address@hidden
> > http://lists.nongnu.org/mailman/listinfo/ltib
> 
> 
> 
> _______________________________________________
> LTIB home page: http://bitshrine.org
> 
> Ltib mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/ltib