Re: AC_C_BIGENDIAN answers "universal" on powerpc-aix

[Ralf Wildenhues]

* Peter O'Gorman wrote on Thu, Aug 14, 2008 at 08:04:20PM CEST:
> On Thu, Aug 14, 2008 at 07:16:58PM +0200, Ralf Wildenhues wrote:
> > (Hereby volunteering to do my part in it in the next 48 hours,
> > but please others do test as well if possible).
> 
> I will test on aix, i386-darwin, and i386-linux

Thanks.

> > This should have a NEWS entry, like
> > 
> > ** AC_C_BIGENDIAN does not mistakenly report "universal" for some
> >    bigendian hosts, a regression introduced with universal binary
> >    support in 2.62.
> 
> Wasn't it 2.61?

I don't think so.  The NEWS entry is for 2.62.

> > > + # Check for potential -arch flags it is not universal unless
> > > + # there are some -arch flags, note that *ppc* also matches
> > > + # ppc64. This check is also rather less than ideal.
> > > + case "${CC} ${CFLAGS} ${CPPFLAGS} ${LDFLAGS}" in  #(
> > > +   *"ppc"*|*"i386"*|*"x86_64"*) ac_cv_c_bigendian=universal;;
> > 
> > Why not check for
> >   *-arch\ ppc* | *-arch\ i386* | *-arch\ x86_64*
> 
> Was worried about whitespace between the -arch and ppc for example.

Then check for *-arch*ppc* etc.?

> I suppose I could do something like:
> set _universal_dummy ${CC} ${CFLAGS} ${CPPFLAGS} ${LDFLAGS}
> shift
> _universal_flags=`echo address@hidden
> to collapse the whitespace, but it seems unnecessary, we have already
> restricted it to the Apple compiler, so it is likely that if the strings
> ppc, i386, x86_64 appear then they are for -arch flags.

Well is there any change for cross compilation, where $CC contains the
target name strings?

I guess the danger is rather low in practice.

> The intention here is to limit the change to the Apple compiler and only
> then when it seems likely that a multi-arch build is occuring.

I think it was feared that other systems also got the "great" idea of
having some kind of universal binaries...

Cheers,
Ralf