tests: four guards that do not hold what they say: a .pyc the CLI reference's child writes under templates/, a launcher guard blind to family.py, a quoting-sentence inventory with three holes, and an S5778 pytest.raises #665

Goal: each guard fails when the thing it names is broken, and the CLI reference's children leave no bytecode in the templates.

Filed by lane openRepoShape-5 on Brett Heap's word of 2026-10-16T12:17:02Z ("#461's test/CI/doc lines (~27)"), from four lines on #461: two "Tests (#554's review)" lines in its body (body, the .pyc one), lane 5's own line from #578 (comment) and two of lane 4's (R588, the inventory test and the S5778 line).

Lands after: nothing is needed. If Brett rules the Git Bash quoting dialect (#532's decision, checklist line on #459) before this starts, fold the inventory line into that change, because the ruling moves NAMED and the sentence the test reads.

Lines (path:line at main 879d8047)

• tests/test_cli_reference.py leaves templates/family-root/scripts/__pycache__/bootstrap.cpython-312.pyc (Tracking: the test suite, CI and the docs #461 body, "Tests (tests: every test that clones a just-pushed local remote by its plain path now clones it with clone_just_pushed, so the macOS failed to copy file race of #440 cannot reach it; load_without_bytecode keeps the doctor's and the Windows-path tests' loads of the family root's scripts from writing .pyc into the templates; the not-a-list doctor cases hold INVALID, exit 1; and GOOD_KEYS holds th #554's review)"). scripts/render-cli-reference.py runs the template's siblings.py --help (PROJECT_TOOLS, the row at :201-206; capture_help, :382-396) in a child whose child_env() (:268-296) sets no PYTHONDONTWRITEBYTECODE, and templates/family-root/scripts/siblings.py:196 does from bootstrap import ... from its own directory. Reproduced at d4a7fab9 in a fresh clone: env -u PYTHONDONTWRITEBYTECODE PYTHONPATH=<clone>/scripts python3 templates/family-root/scripts/siblings.py --help left an 11533-byte bootstrap.cpython-312.pyc there, ignored, so git status stays silent. Nothing reads every file under that template after the test today, so the suite is green by order (the same reason as Both roots' scripts/bootstrap.py: main escapes what stdout cannot encode and spells each value it prints, so name: "a\ud83db" or 日本 under PYTHONIOENCODING=ascii no longer ends make bootstrap in a traceback (#391) #456's CI failure for the assembly root). Direction: add "PYTHONDONTWRITEBYTECODE": "1" to child_env()'s env.update block and one sentence to its docstring, so a person running the generator gets the same; a test beside test_child_env_never_lets_a_child_reach_the_network (tests/test_cli_reference.py:467) that child_env() holds it. No --help text changes, so docs/cli.md is not regenerated.
• tests/test_validator_child_encoding.py's launcher guard does not see family.py (lane 5's line from siblings.py and family.py: make siblings asks about the family folder before it clones, clones each member on its tracking branch, and family.py siblings runs the holder's own copy (#543) #578). interpreter_runs (:616-631) looks for subprocess.run([sys.executable, ...]) by ast.unparse(argv.elts[0]) == "sys.executable", and LAUNCHERS (:610-613) lists scripts/bump-leg.py and update-shape.py. scripts/family.py:6470-6474 runs its child as [sys.executable or PYTHON, SIBLINGS_SCRIPT, ...] with env=inherited_child_environment(). I ran the guard's own helper over the three files at 879d8047: bump-leg.py 1, update-shape.py 1, scripts/family.py 0. So the guard would not notice if family.py siblings dropped the child's env= again (adopt-project.py plan raises UnicodeEncodeError in its terminal report, after the plan file is written, for a printable name stdout cannot encode (日本.md under cp1252, as on a redirected Windows stdout), so plan exits 1 before plan written to #214). test_family_py_siblings_escapes_what_its_stdout_cannot_encode (tests/test_family_siblings_placement.py:653) catches that case; the static guard could still list it. Direction: interpreter_runs accepts sys.executable or PYTHON as well, and LAUNCHERS gains "scripts/family.py": "inherited_child_environment". Run the mutant once (drop env= at :6474) and see the guard fail.
• The inventory test has gaps (lane 4's L4-68; "optional hardening", the code is right). test_the_tools_that_quote_for_the_host_are_the_ones_the_sentence_names (tests/test_readme_windows_quoting_sentence.py:106-120) lists the shipped tools that call a host-dialect quoter. R588 found four holes, and I read each in the code: a quoter imported under another name (from repo_shape import shell_line as joined) is missed, because host_calls (:82-103) compares the callee's name with HOST_QUOTERS (:37); a shipped Python file with no .py suffix is not scanned, and the root bootstrap (#!/usr/bin/env python3) is one; it pins files (NAMED, :40), not lines, so a family.py call switched to POSIX quoting would still pass while the README sentence names the file; and all of scripts/repo_shape.py is exempt (:111) where only the bodies of shell_quoted and shell_line need to be. Direction: follow ImportFrom ... as; scan the shebang files that git ls-files lists; pin each tool's calling functions, not the file; exempt the two bodies by name. Run R588's mutants V2, V3 and V5 and see each fail.
• Sonar S5778 at tests/test_bump_leg_paste_lines.py:402-404 (lane 4's L4-69). bump_leg_module.leg_placement(leg) is evaluated inside with pytest.raises(bump_leg_module.LegNotAsFound), so an exception from it would count as the expected one. The gate passes. Direction: build the placement before the with (placement = bump_leg_module.leg_placement(leg)), as the test above it does (:383).

Lane 4's triage. Lane 4 had the last two lines as L4-68 and L4-69, known limits in its T17 that it would leave unfiled, and has dropped every #461 line from its proposal for lane 5 (#473 comment). Brett Heap's word of 2026-10-16T12:17:02Z covers them here; if he would rather leave them as known limits, drop them and keep the first two lines.

Overlaps. Open PRs: none of the seven open PRs (#621, #627, #628, #630, #634, #639, #656) touches these files; #630 changes README.md at :152-158, where the sentence the inventory test reads sits, but not the test. Lane 5's in-flight branches: fix/647-branch-left-bump-leg-setup edits tests/test_bump_leg_paste_lines.py at :53, :228 and :232 (the S5778 hunk is :402-404, a different place in the same file).

Mark each line → this issue on #461, and tick it there when this closes.

🤖 Generated with Claude Code