Re: unreliable detection of universal builds in AC_C_BIGENDIAN

[Peter O'Gorman]

On Mon, Dec 29, 2008 at 10:51:40PM +0100, Ralf Wildenhues wrote:
> Hello everyone,
> 
> * Eric Blake wrote on Fri, Dec 26, 2008 at 07:13:19PM CET:
> > According to Bruno Haible on 12/26/2008 4:51 AM:
> > > No it wouldn't. $CC, $CFLAGS etc. can also use tabs instead of spaces as 
> > > word
> > > separator. To iterate over the words of a command, the simplest way is 
> > > really
> > > a 'for' loop.
> 
> > How likely is it that -arch appears multiple times, but always with the
> > same argument (or asked another way, my compact version appears immune to
> > false negatives, but how likely is it to trigger false positives that only
> > your more verbose for loop would detect)?
> 
> FWIW, in my humble opinion, verbosity of configure code should not
> prevent "more correct" functioning here.  IOW, I like Bruno's version
> better here; if only because the current version of the code already
> provoked confusion with users, at least Bruno.
> 
> (As a less relevant aside, Bruno's version is probably faster, too.)
> 
> Just my 2 cents.

Bruno's patch looks fine to me too.

Peter
-- 
Peter O'Gorman
address@hidden