[oe] mips site files update

[Tom Rini]

On Sun, Dec 07, 2008 at 01:34:00PM -0800, Khem Raj wrote:
> On (07/12/08 14:18), Tom Rini wrote:
> > On Sun, Dec 07, 2008 at 12:54:30PM -0800, Khem Raj wrote:
> > > On (06/12/08 17:50), Tom Rini wrote:
> > > > On Sat, Dec 06, 2008 at 04:10:50PM -0800, Khem Raj wrote:
> > > > 
> > > > > Hi 
> > > > > 
> > > > > Trying to build for mips based machines. I needed this patch to get the
> > > > > console-image building. 
> > > > > 
> > > > > OK for .dev
> > > > 
> > > > Can we please please please start splitting things up into common-linux
> > > > and common-glibc/uclibc instead?  That way we fix more than mips.
> > > 
> > > Good. I figured that I only needed this change to get it going.
> > > 
> > > Thanks
> > > -Khem
> > 
> > > diff --git a/site/common-linux b/site/common-linux
> > > index 8d915d0..17d7a8d 100644
> > > --- a/site/common-linux
> > > +++ b/site/common-linux
> > > @@ -10,3 +10,7 @@ samba_cv_HAVE_IRIX_SPECIFIC_CAPABILITIES=${samba_cv_HAVE_IRIX_SPECIFIC_CAPABILIT
> > >  # coreutils
> > >  fu_cv_sys_stat_statfs2_bsize=${fu_cv_sys_stat_statfs2_bsize=yes}
> > >  
> > > +# popt
> > > +ac_cv_va_copy=${ac_cv_va_copy=yes}
> > > +ac_cv___va_copy=${ac_cv___va_copy=yes}
> > > +
> > 
> > I've been thinking about this (but not found much time to act, sorry),
> > but we should, at least for common-linux/glibc/uclibc stop grouping by
> > package and start grouping by function.  And in this case, perhaps we
> > should do:
> 
> Configure tests may differ for same function on different packages. Its
> useful to know which package needs what defines. va_copy may be tested
> in different ways in different package's configure scripts.

True.  But we don't list every package which wants va_copy vs __va_copy,
just the first one someone hit.  IOW, it could be interesting
information, but except for packages which do MYPACKAGE_ac_cv_foofunc,
we don't know anything but the first one someone hit.  I wouldn't object
to listing the packages, but I think grouping functions together makes
more sense than grouping packages together.

> > # va_copy
> > ac_cv_va_copy=${ac_cv_va_copy=yes}
> > ac_cv___va_copy=${ac_cv___va_copy=yes}
> 
> > 
> > And similar for *{get,set}*uid* and so on.
> > 
> > In both -glibc and -uclibc, as that's a glibc feature.
> 
> Thats why a common file is preferred most of defines could be common
> for uclibc and glibc. Keeping two copies is unnessary. 

Yes, that too.

> Anyway for now I will add this to both -uclibc and -glibc files.

Thanks.

> How they should be organised logically I will leave for laters.

I'll start a thread / comment on my bug sometime tonight hopefully, but
common-libc? :)

-- 
Tom Rini