Re: AC_C_PROTOTYPES and CC=g++

[Akim Demaille]

| When CC=g++ is used, AC_C_PROTOTYPES seems to get the wrong answer.  I
| think this is due to _AC_PROG_CC_STDC needing K&R style function
| definitions to work before declaring the compiler ANSI capable.
| 
| I know CC=g++ is not really sensible, but there's some setups to make
| "exit" work with it, so maybe it's a semi-supported feature.  In any
| case, unless there's a particular reason for _AC_PROG_CC_STDC to mix
| ANSI and K&R I thought to suggest:
| 
|         * lib/autoconf/c.m4 (_AC_PROG_CC_STDC): Don't require K&R definitions
|         to work, only ANSI, for the benefit of CC=g++ or similar.
| 
| 
| Index: c.m4
| ===================================================================
| RCS file: /cvsroot/autoconf/autoconf/lib/autoconf/c.m4,v
| retrieving revision 1.160
| diff -u -r1.160 c.m4
| --- c.m4      2 Mar 2002 15:19:48 -0000       1.160
| +++ c.m4      25 Mar 2002 22:40:31 -0000
| @@ -752,9 +752,7 @@
|  /* Most of the following tests are stolen from RCS 5.7's src/conf.sh.  */
|  struct buf { int x; };
|  FILE * (*rcsopen) (struct buf *, struct stat *, int);
| -static char *e (p, i)
| -     char **p;
| -     int i;
| +static char *e (char **p, int i)
|  {
|    return p[i];
|  }

I don't think this is the right means to do it.  A proper ANSI C
compiler still accepts K&R code.  So I'd suggest that you wrap this
with __cplusplus and probably __STDC__ too.