Index: src/string/api.c
===================================================================
--- src/string/api.c	(revision 39517)
+++ src/string/api.c	(working copy)
@@ -410,6 +410,35 @@
 
 */
 
+static int
+destination_aux(ARGIN(const STRING *a), ARGIN(const STRING *b),
+    ARGOUT(const CHARSET **c), ARGOUT(const ENCODING **e))
+{
+    if (a->encoding == b->encoding && a->charset == b->charset) {
+        *c = a ->charset;
+        *e = a->encoding;
+        return 1;
+    }
+    if (a->encoding == Parrot_utf8_encoding_ptr && b->encoding == Parrot_fixed_8_encoding_ptr) {
+        *c = a ->charset;
+        *e = a->encoding;
+        return 1;
+    }
+    return 0;
+}
+
+static void
+destination(ARGIN(const STRING *a), ARGIN(const STRING *b),
+    ARGOUT(const CHARSET **c), ARGOUT(const ENCODING **e))
+{
+    if (destination_aux(a, b, c, e))
+        return;
+    if (destination_aux(b, a, c, e))
+        return;
+    *c = Parrot_unicode_charset_ptr;
+    *e = Parrot_utf16_encoding_ptr;
+}
+
 PARROT_EXPORT
 PARROT_WARN_UNUSED_RESULT
 PARROT_CAN_RETURN_NULL
@@ -425,18 +454,14 @@
 
     /* a table could possibly simplify the logic */
     if (a->encoding == Parrot_utf8_encoding_ptr &&
-            (b->charset == Parrot_ascii_charset_ptr ||
-             b->charset == Parrot_iso_8859_1_charset_ptr)) {
-        if (a->strlen == a->bufused) {
-            *e = Parrot_fixed_8_encoding_ptr;
-            return Parrot_ascii_charset_ptr;
-        }
+            (b->charset == Parrot_ascii_charset_ptr /*||
+             b->charset == Parrot_iso_8859_1_charset_ptr*/)) {
         *e = a->encoding;
         return a->charset;
     }
     if (b->encoding == Parrot_utf8_encoding_ptr &&
-            (a->charset == Parrot_ascii_charset_ptr ||
-             a->charset == Parrot_iso_8859_1_charset_ptr)) {
+            (a->charset == Parrot_ascii_charset_ptr /*||
+             a->charset == Parrot_iso_8859_1_charset_ptr*/)) {
         if (b->strlen == b->bufused) {
             *e = Parrot_fixed_8_encoding_ptr;
             return a->charset;
@@ -561,14 +586,17 @@
         a->encoding = enc;
     }
     else {
-        /* upgrade to utf16 */
-        Parrot_utf16_encoding_ptr->to_encoding(interp, a, NULL);
-        b = Parrot_utf16_encoding_ptr->to_encoding(interp, b,
-                Parrot_gc_new_string_header(interp, 0));
-
-        /* result could be mixed ucs2 / utf16 */
-        if (b->encoding == Parrot_utf16_encoding_ptr)
-            a->encoding = Parrot_utf16_encoding_ptr;
+        enc = NULL;
+        destination(a, b, &cs, &enc);
+        /* upgrade to chosen destination */
+	if (cs != a->charset)
+	    a = cs->to_charset(interp, a, NULL);
+	if (enc != a->encoding)
+            a = enc->to_encoding(interp, a, NULL);
+	if (cs != b->charset)
+	    b = cs->to_charset(interp, b, NULL);
+	if (enc != b->encoding)
+            b = enc->to_encoding(interp, b, NULL);
     }
 
     /* calc usable and total bytes */