Re: coreutils 5.0 on Tru64 Unix

[Jim Meyering]

Tony Leneis <address@hidden> wrote:
>       Tru64 Unix declares __sys_siglist in signal.h.  It works just like
> _sys_siglist and sys_siglist on other systems.  I'm not sure what the
> rationale was behind it, but it would be nice if configure could detect
> this and use the same work-around in kill.c that _sys_siglist does.  (It's
> necessary because Tru64 Unix doesn't have strsignal().)
>
>       Assuming you want to follow the same naming scheme, I guess config.h
> would set HAVE_DECL___SYS_SIGLIST to indicate there's a __sys_siglist.  Here
> is the patch for src/kill.c:
>
> --- kill.c      2002-09-02 00:29:44.000000000 -0700
> +++ patchedkill.c       2003-04-04 22:20:43.000000000 -0800
> @@ -52,6 +52,9 @@
>  #  if HAVE_DECL__SYS_SIGLIST || defined _sys_siglist
>  #   define sys_siglist _sys_siglist
>  #  endif
> +#  if HAVE_DECL___SYS_SIGLIST || defined __sys_siglist
> +#   define sys_siglist __sys_siglist
> +#  endif
>  # endif
>  # if HAVE_DECL_SYS_SIGLIST || defined sys_siglist
>  #  define strsignal(signum) (0 <= (signum) && (signum) <= SIGNUM_BOUND \

I've applied that, but used #elif rather than a separate #if block.

> I'm not an autoconf person, but here's a test program for __sig_siglist:

Thanks.  Here's the change I've checked in:

Index: configure.ac
===================================================================
RCS file: /fetish/cu/configure.ac,v
retrieving revision 1.25
retrieving revision 1.26
diff -u -p -u -r1.25 -r1.26
--- configure.ac        5 Apr 2003 15:44:29 -0000       1.25
+++ configure.ac        5 Apr 2003 15:45:55 -0000       1.26
@@ -250,7 +250,7 @@ if test $jm_cv_sys_tiocgwinsz_needs_term
 fi
 
 # For src/kill.c.
-AC_CHECK_DECLS([strsignal, strtoimax, sys_siglist, _sys_siglist])
+AC_CHECK_DECLS([strsignal, strtoimax, sys_siglist, _sys_siglist, 
__sys_siglist])
 
 jm_LIB_CHECK