Clarify default encoding behavior of open() with UTF-8 mode PEP 686 (Windows confusion)

[botkero]

Documentation

Hi,

The documentation for open() currently states:

In text mode, if encoding is not specified the encoding used is platform-dependent: locale.getencoding() is called to get the current locale encoding.

However, my observations on Windows suggest that the actual behaviour is unclear or at least confusing in practice.

with open("result_utf-8.txt", "w") as f:
    f.write("fff")

with open("result_ansi.txt", "w") as f:
    f.write("ffföäü")

Started with python -B myscript.py

I use Python 3.13.1 on Windows 11 (German language pack).

---

According to debugging, the encoding used internally appears to be cp1252 (matching locale.getencoding()).
However, when opening the files in Notepad:

• result_utf8.txt is detected as UTF-8
• result_ansi.txt is detected as ANSI (cp1252)

This gives the impression that the effective encoding depends on file content, which is unexpected.

I would expect one of the following:

• The encoding is consistently cp1252 (as per locale.getencoding()), or
• The encoding is consistently UTF-8 (e.g. due to UTF-8 mode / PEP686)

---

I'm not sure exactly what category this falls under, whether it's documentation or a bug in Python. Maybe I'm completely off track here, and actually, everything fits. Please clarify this for me.

Edit:

Hmm, after further investigation. It seems everything is correct after all. I should have checked the bytes. The result is ANSI in both cases. Notepad just incorrectly displays UTF-8 in the first case.