tests(perf[conftest]) Pin $SHELL=/bin/sh — 51% faster suite

[tony]

Summary

Add a 14-line autouse fixture to the root conftest.py that pins $SHELL=/bin/sh for every test. Drops uv run py.test wall time by ~51% on this machine — from ~76s to ~36s — by bypassing the contributor's interactive shell init (zsh + oh-my-zsh + plugins, etc.) on every tmux pane the suite spawns.

Why this works

tmux falls back to $SHELL for new panes when the default-shell option is unset. libtmux's bundled test .tmux.conf does not set default-shell, so every pane in tmuxp's test suite spawns the contributor's interactive $SHELL — typically /bin/zsh — and pays the full rcfile cost. With hundreds of pane spawns across the workspace/builder tests, that's the single largest line item in suite wall time.

monkeypatch.setenv("SHELL", "/bin/sh") in a function-scoped autouse fixture forces every test pane to spawn /bin/sh instead. Empty rcfiles, deterministic prompt, no plugin chains. monkeypatch restores the prior value at teardown so no global state leaks.

Measured timing (this machine, uv run py.test)

master                ~76-78s pytest, ~71s wall
this branch (run 1)    37.01s pytest, 35.08s wall
this branch (run 2)    36.24s pytest, 34.29s wall

Both runs: 797 passed, 2 skipped, zero reruns. Master had occasional reruns on capture-pane assertions that depended on prompt timing; pinning the shell removes that source of flake along with the perf cost.

Relationship to libtmux PR #662

This same fix exists upstream in tmux-python/libtmux#662 (open since 2026-04-29), which pins $SHELL=/bin/sh in libtmux's own pytest plugin and reports a similar "~70s → ~32s" delta on tmuxp's suite. Once libtmux #662 merges and ships in a release tmuxp picks up, this commit becomes redundant and can be reverted cleanly (the autouse fixture would just be a no-op repeating what libtmux's plugin already does).

Landing this at the tmuxp conftest level lands the gain immediately rather than waiting on the upstream PR.

What's not in this PR (deliberately)

• The USING_ZSH constant and zshrc autouse fixture become dead at runtime once $SHELL is always /bin/sh (no zsh first-run greeting to suppress, no rcfile chain to write a quieting .zshrc for). They're left in place to keep this PR's diff to the single load-bearing change. Cleanup belongs in a follow-up — easier to review and easier to revert if the perf claim ever needs to be questioned.

Test plan

• uv run ruff check . --fix --show-fixes — clean
• uv run ruff format . — clean
• uv run mypy — clean (124 source files)
• uv run py.test — 797 passed, 2 skipped, no reruns (both runs)
• just build-docs — succeeds
• Timing comparison vs master — see table above

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.98%. Comparing base (c675892) to head (a4b416b).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1041      +/-   ##
==========================================
+ Coverage   81.96%   81.98%   +0.02%     
==========================================
  Files          28       28              
  Lines        2545     2548       +3     
  Branches      485      485              
==========================================
+ Hits         2086     2089       +3     
  Misses        328      328              
  Partials      131      131              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[tony]

Code review

No issues found. Checked for bugs and CLAUDE.md compliance.

Claude Code generated.