Re: avoiding obsolete macros like AC_TRY_RUN

[Bruno Haible]

Ralf Wildenhues wrote:
> diff --git a/m4/getloadavg.m4 b/m4/getloadavg.m4
> index 28237d7..e5ea9a4 100644
> --- a/m4/getloadavg.m4
> +++ b/m4/getloadavg.m4
> @@ -174,14 +174,14 @@ AC_CHECK_HEADERS([nlist.h],
>                             `n_un' member.  Obsolete, depend on
>                             `HAVE_STRUCT_NLIST_N_UN_N_NAME])], [],
>                 address@hidden:@include <nlist.h>])
> - AC_TRY_LINK([#include <nlist.h>],
> -             [struct nlist x;
> -              #ifdef HAVE_STRUCT_NLIST_N_UN_N_NAME
> -              x.n_un.n_name = "";
> -              #else
> -              x.n_name = "";
> -              #endif],
> -             [AC_DEFINE([N_NAME_POINTER], [1],
> -                        [Define to 1 if the nlist n_name member is a 
> pointer])])
> + AC_LINK_IFELSE([AC_LANG_PROGRAM([[#include <nlist.h>]],
> +                [[struct nlist x;
> +                 #ifdef HAVE_STRUCT_NLIST_N_UN_N_NAME
> +                 x.n_un.n_name = "";
> +                 #else
> +                 x.n_name = "";
> +                 #endif]])],
> +                [AC_DEFINE([N_NAME_POINTER], [1],
> +                           [Define to 1 if the nlist n_name member is a 
> pointer])])
>  ])dnl
>  ])# gl_PREREQ_GETLOADAVG

I wonder about the indentation here: The [AC_DEFINE line is indented by 3 more 
spaces
than the [AC_LANG_PROGRAM argument, making it look like the AC_DEFINE 
expression was
an argument to AC_LANG_PROGRAM, not an argument to AC_LINK_IFELSE.

The indentation in m4/fsusage.m4 and m4/ls-mntd-fs.m4 is also bizarre.

Bruno