On Sun, Jun 12, 2016, at 12:50, Steven D'Aprano wrote: > I think Windows also gets it almost write: NTFS uses UTF-16, and (I > think) only allow valid Unicode file names.
Nope. Windows allows any sequence of 16-bit units (except for a dozen or so ASCII characters) in filenames. Of course, you're not particularly _likely_ to encounter invalid surrogates, since nothing is going to create them without deliberately setting out to (unlike Linux where 'invalid' filenames will be created by any program using the 'wrong' locale). -- https://mail.python.org/mailman/listinfo/python-list