Re: AC_SUBST and AC_CACHE_VAL

[Ralf Wildenhues]

* Stefan Bienert wrote on Wed, May 13, 2009 at 10:33:06PM CEST:
> > What does the last line have to do with the ones before that?
> 
> Oh, I forgot to mention: I have this stuff in a macro. It's meant to
> check for certain flags and if accepted by the program, the flag is
> added to the flag-list. Now I have another macro, which invokes the flag
> checks by AC_REQUIRE... If someone invokes the macro at another place,
> with the AC_SUBST in the end, the concatenation happens.

OK, but still the last line had no logical connection with the other
ones, so I wondered whether we were missing something.

> > If an initial value of SB_FOOS from the user (./configure SB_FOOS=...)
> > is to be honored, then maybe you should just check for presence of bar
> > before adding it?
> >   case " $SB_FOOS " in
> >   *\ bar\ *) ;;
> >   *) SB_FOOS="$SB_FOOS bar";;
> >   esac
> 
> Never thought of that before! Many thanks, I will check my setups where
> this applies. Tres cool, I would say! But why is nobody using AS_CASE?

Old habit.  Use AS_CASE, please.  It makes a difference only if its
arguments expand AC_REQUIRE statements.

> > If the user is not involved here, then how come you save and restore
> > SB_FOOS but not initialize it early in configure?
> 
> Isn't SB_FOOS empty if not initialised?

Sure, but do you want it to be seedable or not?

Cheers,
Ralf