Today I worked on this bug, creating a test case, to finally discover that I cannot reproduce the bug in my computer.
There are some warnings about wide characters being printed, so it feels like it's working by chance. Also, the patch by Gregor doesn't seem correct as it makes a lot of noise in other tests, and many fail. After some investigation, I came to the conclusion that toString should NOT be touched, as strings are currently correctly utf8 encoded, so it only makes sense to do the bit-juggling when outputting to a file/filedescriptor. Trimming the patch to do that makes all tests pass, and suppresses the wide char warnings. Also, it should be noted that CPAN#27793 and CPAN#14579 seem to be related to this. Upstream looks a little dead to me. I'm attaching my test case. -- Martín Ferrari
encodings.t
Description: Troff document
