Re: 2.59d on OpenServer

[Paul Eggert]

Tim Rice <address@hidden> writes:

> OSR5 didn't have /usr/bin/posix/sh so there wasn't
> and "backward compatability" reasson to use the older ksh.

This is a regression, then, and it was caused by the patch to m4sh.m4
described here:

http://lists.gnu.org/archive/html/autoconf-patches/2006-03/msg00081.html

Given the unanticipated problems on OpenServer, perhaps we should go
back to the proposal (0a) in that message.  Does the following patch
fix things for you?  (I haven't installed it.)

2006-06-06  Paul Eggert  <address@hidden>

        * lib/m4sugar/m4sh.m4 (AS_BOURNE_COMPATIBLE): Don't set BIN_SH, since
        we want to select a shell that doesn't need BIN_SH to be set to fix
        things.  See (0a) under:
        http://lists.gnu.org/archive/html/autoconf-patches/2006-03/msg00081.html
        (_AS_DETECT_BETTER_SHELL): Don't prepend /usr/bin/posix to PATH.
        This fixes a bug reported by Tim Rice for OpenServer 6.0
        <http://lists.gnu.org/archive/html/bug-autoconf/2006-06/msg00013.html>
        where /usr/bin/posix/sh has a too-small limit for string literals.

--- lib/m4sugar/m4sh.m4 31 May 2006 09:44:39 -0000      1.190
+++ lib/m4sugar/m4sh.m4 6 Jun 2006 22:51:51 -0000
@@ -172,7 +172,6 @@ if test -n "${ZSH_VERSION+set}" && (emul
 else
   case `(set -o) 2>/dev/null` in *posix*) set -o posix;; esac
 fi
-BIN_SH=xpg4; export BIN_SH # for Tru64
 DUALCASE=1; export DUALCASE # for MKS sh
 ])
 
@@ -228,7 +227,7 @@ if test "x$CONFIG_SHELL" = x; then
         _AS_RUN([_AS_DETECT_SUGGESTED_BODY]) 2> /dev/null],
     [],
     [as_candidate_shells=
-    
_AS_PATH_WALK([/usr/bin/posix$PATH_SEPARATOR/bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH],
+    _AS_PATH_WALK([/bin$PATH_SEPARATOR/usr/bin$PATH_SEPARATOR$PATH],
       [case $as_dir in
         /*)
           for as_base in sh bash ksh sh5; do