[commit] Avoid CATOBJEXT=NONE; Was: [patch/rfc] Only use ../intl/when present

[Andrew Cagney <cagney at gnu dot org>]

Andrew Cagney wrote:
> Andrew Cagney wrote:
>
> > Hello,
> >
> > This patch should stop getgext.m4 trying to use ../intl when it isn't 
> > present.
> >
> > I can't actually test it - all the systems I looked to test it on had 
> > a bundled gettext :-/

The attached should fix a further problem - it was appending NONE as the 
catalog suffix.  Again tested on a system with installed gettext, and 
with --disable-nls (which isn't very useful).

If people still encounter problems can they please include a few details 
such as the broken Makefile.  Otherwize I'm put in a situtation where 
I'm flying blind :-(

committed, configures also regenerated,
Andrew

2005-01-27  Andrew Cagney  <cagney@gnu.org>

	* gettext.m4: Don't use NONE as a default for CATOBJEXT.

Index: gettext.m4
===================================================================
RCS file: /cvs/src/src/gettext.m4,v
retrieving revision 1.4
diff -p -u -r1.4 gettext.m4
--- gettext.m4	24 Jan 2005 20:00:49 -0000	1.4
+++ gettext.m4	27 Jan 2005 14:44:24 -0000
@@ -40,7 +40,7 @@ AC_DEFUN([CY_WITH_NLS],
 	dnl catgets is only used if permitted by option --with-catgets.
 	nls_cv_header_intl=
 	nls_cv_header_libgt=
-	CATOBJEXT=NONE
+	CATOBJEXT=
 
 	AC_CHECK_HEADER(libintl.h,
 	  [AC_CACHE_CHECK([for gettext in libc], gt_cv_func_gettext_libc,
@@ -81,7 +81,7 @@ AC_DEFUN([CY_WITH_NLS],
 	dnl In the standard gettext, we would now check for catgets.
         dnl However, we never want to use catgets for our releases.
 
-        if test "$CATOBJEXT" = "NONE" && test -d $srcdir/../intl; then
+        if test x"$CATOBJEXT" = x && test -d $srcdir/../intl; then
 	  # Neither gettext nor catgets in included in the C library.
 	  # Fall back on GNU gettext library (assuming it is present).
 	  nls_cv_use_gnu_gettext=yes