Hi, t/mro/package_aliases_utf8.t is failing for me: # Failed test 11 - replacing nonexistent nested packages by assigning a glob to a glob updates isa caches (인ንʵ) at ./test.pl line 864 # got "Cannot decode string with wide characters at ../lib/Encode.pm line 176." # expected "ok 1\nok 2\nok 3\nok 4" This happens because the test runs `perl foo.pl $unicode_string` and foo.pl blindly tries to decode @ARGV as UTF-8. This fails because I've got PERL_UNICODE=A in my environment (more specifically, PERL_UNICODE=SAL and LANG=en_US.UTF-8), so @ARGV gets automatically decoded by perl. By the time foo.pl calls Encode::decode("UTF-8", $ARGV[0]), it's already unicode text. The solution is probably to set $ENV{PERL_UNICODE} = 0 before shelling out.