Re: A simplification of _AC_CANONICAL_SPLIT

[Stepan Kasal]

Hello,

On Mon, Aug 22, 2005 at 03:21:36PM -0700, Paul Eggert wrote:
> > +case $ac_cv_$1 in
> > +  -*) false ;;
> > +  *-*-*) : ;;
> > +  *) false ;;
> > +esac || AC_MSG_ERROR([invalid value of canonical $1])
> 
> Sorry, I still don't get it.  Why is that first "false" needed?

(I agree this thread is getting too long, so we are all tired.)

I want to use this:
 IFS=-
 set $ac_cv_$1
 AC_SUBST([$1_cpu], [$[1]])dnl
 AC_SUBST([$1_vendor], [$[2]])dnl
 shift; shift
 AC_SUBST([$1_os], [$[*]])dnl

If $ac_cv_$1 started with "-", set would misinterpret it as option.
And if it had less than 3 fields, the shifts would complain.

Since both of these are invalid anyway, I want to filter them out.

Regards,
        Stepan