1 test fails about natsort HOT 5 CLOSED

I cannot reproduce. I'm not even sure how an OSError could be thrown here. I wonder if it is something OS-specific.

Can you do the following and let me know what happens?

$ python3.9 -c "import unicodedata; print(unicodedata.numeric('Å'))"

from natsort.

$ python3.9 -c "import unicodedata; print(unicodedata.numeric('Å'))"
Traceback (most recent call last):
  File "<string>", line 1, in <module>
ValueError: not a numeric character

from natsort.

Bummer - that's what I get too. Not an OSError. I'm going to see if I can add BSD to my CI jobs and hopefully that exposes something.

from natsort.

Oh my goodness... I can replicate: https://github.com/SethMMorton/natsort/actions/runs/4300958649/jobs/7497664518

I'll solve this tomorrow.

BUT, based on which tests are failing this clearly has something to do with the locale library, which is interesting because I did not touch that in the last update.

from natsort.

I can trace this to locale.strxfrm. https://github.com/SethMMorton/natsort/actions/runs/4310592727/jobs/7519180370#step:3:1715

It fails with an OSError when given 'Å'. This is not the first time I have encountered a mind-boggling bug with locale on FreeBSD - sigh. I guess I'll have to add more code to natsort to work around bugs in the locale library with BSD.

from natsort.