locale-encoding

Locale encoding and the Windows CI cell

The windows-latest CI runner is an English Windows image, so locale.getencoding() there is cp1252, which cannot represent Cyrillic at all. The primary development machine is a Russian Windows install with cp1251, which can. Ubuntu cells are UTF-8.

This makes a whole class of bugs invisible locally: anything that falls back to the locale encoding works on cp1251 and on Linux, and fails only on the CI Windows cell. "It passes on my Windows machine" is therefore not evidence.

It has already happened once. The CLI wrote its result with sys.stdout's locale encoding while --file/--output were pinned to UTF-8, so redirecting CLI output on cp1252 died with UnicodeEncodeError and every CLI test failed on windows-latest only.

Rules

• Always pass encoding="utf-8" explicitly to open(), read_text(), write_text(), and to subprocess.run() when capturing text. Never rely on text=True alone — it decodes with the locale encoding.
• __main__.py calls _use_utf8_stdio() first thing in main(); keep any new stream it touches UTF-8 too.
• To reproduce the CI environment on any platform, force the child's encoding: subprocess.run(..., env=dict(os.environ, PYTHONIOENCODING="cp1252")). See test_cli_emits_utf8_under_a_non_cyrillic_locale_encoding in tests/test_api.py.
• To find latent cases, run the suite with PYTHONWARNDEFAULTENCODING=1 and look for EncodingWarning outside site-packages.