Tweak the definition of the bit complement on UTF-8 data:

if none of the characters in the string are > 0xff,
the result is a complemented byte string, not a (UTF-8)
char string.  Based on the summary in

Subject: Re: [ID 20000918.005] ~ on wide chars
Message-ID: <jSDD6gzkgi/T092yn@efn.org>

This should give us the maximum backward (pre-char string)
compatibility and utf8 compatibility.  The other alternative
would be to limit the bit complement to be always byte only,
taking the least significant byte of the chars.

p4raw-id: //depot/perl@7665

1 files changed, 1 insertions, 1 deletions

diff --git a/doop.c b/doop.c
index 9fd7dfa49f..3d22eb41fe 100644
--- a/doop.c
+++ b/doop.c
@@ -926,7 +926,7 @@ Perl_do_vop(pTHX_ I32 optype, SV *sv, SV *left, SV *right)
 
     if (left_utf && !right_utf)
 	sv_utf8_upgrade(right);
-    if (!left_utf && right_utf)
+    else if (!left_utf && right_utf)
 	sv_utf8_upgrade(left);
 
     if (sv != left || (optype != OP_BIT_AND && !SvOK(sv) && !SvGMAGICAL(sv)))