Thank you for digging into this. I also learned that 5.14.3 is not officially supported, so I am inclined not to worry about the bug and the test failure.

I would like to have zero test failures, however, so I am thinking I will change the use open() call to a use if ... statement so it doesn't load under the broken perl/OS combo.

I would like to have zero test failures, however, so I am thinking I will change the use open() call to a use if ... statement so it doesn't load under the broken perl/OS combo.

Naturally you're free to do what you think is best

but

why do you even care that STDIN/STDERR/STDOUT are in utf8?

most windows machines don't show unicode by default unless you do chcp

test suite shouldn't much care if STD* are utf8, if there is an encoding problem in a failing test, byte information is what is important , not correct drawing/rendering in the local shell

but if you care just binmode $_, q{:utf8} for *STDIN, *STDOUT, *STDERR;

if there is an encoding problem in a failing test, byte information is what is important , not correct drawing/rendering in the local shell

Quite so. I just didn't like the warning about printing a wide character.

Thanks for your response.

The way forward always starts with a minimal test.