Thank you for addressing it. I guess that your change would resolve the reported error. I checked that a non-ascii working directory test with cp1250 locale passed after merging your change.

Unfortunately, even after your change, I realized that test_pip_install_successful and some other tests failed on another default encoding like cp932. subprocess.check_output called from run_pip raised an exception during decoding output of pip, which is like

Processing c:\users\hugov\あ\poetry\tests\fixtures\distributions\demo-0.1.0-py2.py3-none-any.whl
Installing collected packages: demo
Successfully installed demo-0.1.0

There the parent assumed the default cp932 while pip returned the output in utf-8. So it may be about the behavior of pip. I guess I or any other users will raise an issue if this would be an apparent problem. (At least I know that pip doesn't always use utf-8 for output. The reason it passed in CI was because bytes("ä", encoding="utf-8") is valid also in the cp1250 context but bytes("あ", encoding="utf-8") isn't in the cp932 one.)

pip_install() mostly isn't very interesting to poetry any more: it is almost - though sadly not quite entirely - deprecated in favour of the "modern" installer.

Further fixes welcome I expect, meanwhile if we're happy that this one is good and does fix some things then hopefully we can take that win. Progress is progress.

also none of the callers of pip_install() actually care about the output, so far as I can see: so probably a fix that converts that to use the call variant of Env.run() would be straightforward

This pull request has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.