[Roojool]

What

pr_overlap_encoding errors on Python 3.10 before testing any Unicode handling: its command_output context manager patches subprocess._text_encoding, which that interpreter does not provide. A native Windows run on main produced eight AttributeError errors. The same limitation is recorded in the validation results of #3986, #1764, #1786 and #2391.

Keep the existing CP1252 simulation on newer interpreters. On Python 3.10, intercept the io.TextIOWrapper locale encoding that Popen actually uses. Explicit UTF-8 arguments still pass through unchanged. A real subprocess test checks the exact CP1252-decoded output when no encoding is supplied and exact Unicode output when UTF-8 is supplied. No nonexistent mock attribute is created.

Only tools/tests/test_pr_overlap_encoding.py changes. Production UTF-8 handling from the previous fixes remains unchanged; this repairs the regression tests rather than reimplementing those fixes. No dependency added. Authored by Rujul Talekar with Muse Spark 1.3 and Codex assistance. This is developer test-suite compatibility, with no title or new CI-failure claim.

Tested

Baseline: 2ca6198e9a43424c2417f2c4c000ecd3d141811a. Rebased onto main 436e310ddb56fcf5dd1d40ba1f8113fe8cb8c113 and repeated final validation.

• Native Windows, Python 3.10.11: unchanged baseline python -m unittest tools.tests.test_pr_overlap_encoding -v fails with eight errors, AttributeError: module 'subprocess' has no attribute '_text_encoding'. With the fix, 11/11 pass.
• Linux via WSL2 Ubuntu, Python 3.14.4: the same fixed encoding suite 11/11 passes. The baseline failure concerns Python 3.10; no Linux baseline failure is claimed.
• Final neighboring validation on both systems, 29/29 pass: python -m unittest tools.tests.test_pr_overlap tools.tests.test_pr_overlap_dependencies tools.tests.test_pr_overlap_encoding tools.tests.test_pr_overlap_exports -v (python3 on Linux).
• In a disposable copy outside the repository, removing production's explicit UTF-8 arguments makes the simulation fail on both platforms: Linux raises UnicodeDecodeError; Windows's reader thread raises the decode error and the caller rejects the missing output. Unmodified production decodes exactly on both. The source branch's production file was never altered.
• python tools/check_conventions.py --base upstream/main --head HEAD: passed. git diff --check: passed.
• No full C++ rebuild or title/GPU runs; this changes one Python test file. Tested interpreter versions are 3.10 and 3.14; older interpreters were not checked.

Checklist

• Based on current main; no other open PR implements the same functions
• One topic per PR; follow-ups go in a new PR
• No comments in code except technical debt
• Unimplemented paths throw (NotImplemented_nid_no_patch); silent stubs are listed in TechnicalDebt
• No notes, investigation .md files or images added to the repository (attach them to this PR)
• New third-party code is a submodule built from source
• Generic behaviour, not specific to one title
• Depends on: none
• AI-assisted: yes

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[MarcZX10]

I ran the focused overlap encoding and neighboring regression tests locally on Linux/Python 3.11: 29/29 passed. The conventions check and git diff --check also pass. This exercises Python 3.11’s _text_encoding branch; I did not run Python 3.10 locally. The CPython 3.10 subprocess.Popen implementation sets the default encoding to "locale" before constructing its TextIOWrapper, which matches the fallback branch added here.

[Zurisen]

Native Windows run of e991645c merged into main 2a5eca11 (Windows 11 Home 26200), with Python 3.10.11 and 3.12.10:

• tools/tests/test_pr_overlap_encoding.py: 11/11 on both (on main, 3.10 errors in every case with AttributeError: ... '_text_encoding').
• Guard check: with encoding="utf-8" removed from pr_overlap.gh, it fails on both (test_gh_json_preserves_unicode errors, test_gh_error_preserves_unicode fails), so the regression stays covered on 3.10.

A smaller variant, if you want one, with no version branch or io.TextIOWrapper patch: the fake subprocess.run in command_output can give a text-mode call without an encoding cp1252 itself, which is what a cp1252 default locale does on every version:

def run(command, **kwargs):
    if kwargs.get("text") or kwargs.get("universal_newlines"):
        kwargs.setdefault("encoding", "cp1252")
    return RUN([sys.executable, "-c", code], **kwargs)

With that in place of the _text_encoding patch, the same 10 cases pass on 3.10 and 3.12 and fail the same way without encoding= (commit 2ba0da75 on Zurisen:fix/pr-overlap-test-python310). Your version is fine as is; this is only if a reviewer prefers fewer moving parts.