The real fix for the regexp-related ones is to fix the regexp internal opcodes to be polymorphic with respect to utf8-ness. The way it works right now is not acceptable. I don't know whether I'll get time to work on it any time soon, though. #! perl print "1..4\n"; $_="\xff\xf3"; s/\xff//; print "not " unless $_ eq "\xf3"; print "ok 1\n"; $_="\xff\x{1f3}"; s/\xff//; print "not " unless $_ eq "\xf3"; print "ok 2\n"; { use utf8; $_="\xff\xf3"; s/\xff//; print "not " unless $_ eq "\xf3"; print "ok 3\n"; $_="\xff\x{1f3}"; s/\xff//; print "not " unless $_ eq "\xf3"; print "ok 4\n"; } Hmm. 50% ain't all that good. Fixing this will require that we fix $1 et al as well. --s.Thread Previous | Thread Next