String literals for UTF-8 language names using octal escapes only...

... Leaving absolutely no doubt about the portability.

New Perl script in locale emits the required C++ source when LanguageNames.txt
is updated.

locale/CeePlusPlusifyLanguageNames.pl

@@ -0,0 +1,23 @@
+#!/usr/bin/perl
+# Take LangugeNames.txt and write C++ string literals to standard out,
+# which describe the UTF-8 byte sequence using only the basic coding
+# character set.
+# Copy and paste that over the table of strings in src/Languages.cpp.
+
+open LANGUAGES, "LanguageNames.txt" or die "Can't open LanguageNames.txt: $!\n";
+while( <LANGUAGES> ) {
+    chop;
+    print "\"";
+    @codes = unpack 'C*';
+    @newCodes = ();
+    for $code (@codes) {
+	if ($code <= ord 'z') {
+	    push @newCodes, $code;
+	}
+	else {
+	    push @newCodes, unpack('C*', sprintf("\\%03o", $code));
+	}
+    }
+    print pack 'C*', @newCodes;
+    print "\",\n";
+}