Patch: FYI: PR 5031

[Tom Tromey]

I'm checking this in.

This fixes PR 5031.

setlocale(LC_ALL,"") can return a long form of the locale in some
situations.  This confuses the code that sets user.region.  My fix is
to simply look at LC_MESSAGES only.  I think we have to pick a single
part of the locale since Java and POSIX locales work differently.
LC_MESSAGES seemed like the most reasonable choice to me.

Tom

Index: ChangeLog
from  Tom Tromey  <tromey@redhat.com>
	For PR libgcj/5031:
	* java/lang/natSystem.cc (init_properties): Use LC_MESSAGES to
	choose default locale.

Index: java/lang/natSystem.cc
===================================================================
RCS file: /cvs/gcc/gcc/libjava/java/lang/natSystem.cc,v
retrieving revision 1.42
diff -u -r1.42 natSystem.cc
--- java/lang/natSystem.cc 2002/01/06 21:38:15 1.42
+++ java/lang/natSystem.cc 2002/01/08 21:45:08
@@ -415,7 +415,12 @@
 
   // Set user locale properties based on setlocale()
 #ifdef HAVE_SETLOCALE
-  char *locale = setlocale (LC_ALL, "");
+  // We let the user choose the locale.  However, since Java differs
+  // from POSIX, we arbitrarily pick LC_MESSAGES as determining the
+  // Java locale.  We can't use LC_ALL because it might return a full
+  // list of all the settings.
+  setlocale (LC_ALL, "");
+  char *locale = setlocale (LC_MESSAGES, "");
   if (locale && strlen (locale) >= 2)
     {
       char buf[3];