Re: AC_SEARCH_LIBS optimization

[Noah Misch]

On Tue, Sep 06, 2005 at 05:34:17PM +0200, Stepan Kasal wrote:
> Moreover, I converted the code to use AS_VAR_* macros, so that it
> allows more general application.

AS_VAR_* macros only improve generality where their first argument may or may
not bear a variable reference or command substitution.  This ambiguity only
arises when the first argument contains an M4 argument substitution.

> +  AC_LINK_IFELSE([], [AS_VAR_SET(ac_Search, [$ac_res])])
> +  AS_VAR_SET_IF(ac_Search, [break])dnl

Here and throughout your patch, the first argument is constant, so these uses do
not improve the code.

> --- tests/semantics.at        14 May 2005 07:00:40 -0000      1.50
> +++ tests/semantics.at        6 Sep 2005 15:19:52 -0000
> @@ -83,7 +83,13 @@
>  case "$ac_cv_search_cos" in
>    -loser|-lust) AC_MSG_ERROR([jeez, $ac_cv_search_cos must be a cool 
> library!]) ;;
>  esac
> +])
>  
> +
> +# AC_SEARCH_LIBS (none needed)
> +# ----------------------------
> +AT_CHECK_MACRO([AC_SEARCH_LIBS (none needed)],
> +[
>  AC_SEARCH_LIBS(printf, oser c ust,,
>            [AC_MSG_ERROR([cannot find `printf'])])
>  
> @@ -91,7 +97,6 @@
>    -loser|-lust) AC_MSG_ERROR([jeez, $ac_cv_search_printf must be a cool 
> library!]) ;;
>    -lc) AC_MSG_ERROR([huh, you need to give -lc?])
>  esac
> -
>  ])

This chunk just splits the old test case into two test cases, right?