tests/conftest.py: one helper builds the url.<base>.insteadOf rewrite that clone_just_pushed and read_through_git each build, and the plain-path git submodule update --init sites of #554's review use it (#440) #662

Goal: every test that has git clone a just-pushed local remote reads it through git, and the rule that makes it so is written once.

Filed by lane openRepoShape-5 on Brett Heap's word of 2026-10-16T12:17:02Z ("#461's test/CI/doc lines (~27)"), from three lines on #461: its body (the two "Tests (#554's review)" lines, body) and lane 2's L2-40 (comment).

Lands after: #645's pull request (branch fix/645-fixture-copy-maintenance-race, no pull request yet). It edits tests/conftest.py (the docstring of clone_just_pushed at :85, and a new block after cloned_just_pushed at :137) and turns git's automatic maintenance off for every git the suite starts, the receive-pack of a push included. That may close at its root the race these lines guard against. Read what it landed first, then decide whether the sites below still want the rewrite as a second guard.

Lines (path:line at main 879d8047)

• L2-40, one rule in two places. clone_just_pushed (tests/conftest.py:71-125; the rewrite is built at :112-116 from the RESOLVED remotes directory) and read_through_git (:1395-1415; typed and resolved, as GIT_CONFIG_COUNT entries) each build the url.<file URI>.insteadOf rewrite for a just-pushed local remote. No test fails; a change to one will miss the other. Direction: one helper that returns the prefix list (resolved, and typed on request); clone_just_pushed passes it as -c arguments and read_through_git as GIT_CONFIG_* entries.
• A plain-path git submodule update --init is still git's local clone of a just-pushed remote (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, 2026-10-14)"). Sites: tests/test_family_bump_remove_rollback.py:337 and :908; tests/test_doctor_legs_members_next.py:631 and :987 (the line says :630 and :981; they moved) and :1054, and tests/test_doctor_gitlink_repairs.py:343, which the line does not list. Direction: give each the helper's -c arguments, or a submodule_update_through_git wrapper beside clone_just_pushed. Check each site first: an update over a submodule whose store already exists fetches (upload-pack reads the object store); only an --init of an empty mount is a local clone that walks the remote's files. tests/test_doctor_legs_members_next.py:607 is a deliberate failure probe and stays as it is.
• A hygiene test that refuses a plain-path recursive git clone of a just-pushed remote in tests/ outside conftest.py (optional; same body section). Direction: an AST test, if a rule can separate a clone of the session's remotes (scaffolded["remotes"], remotes / f"{PROJECT}.git") from a clone of a test's own source repository: plain clones of the latter remain, for example tests/test_adopt_submodules.py:150, :263, :272 and tests/test_adopt_ignored_shape_files.py:166, :246. If no rule is that precise, say so on this issue and close the line as not worth a noisy test. After tests: the holder fixtures copy the family fixture with shutil.copytree while git's background gc or maintenance can still be packing it, so a test errors at setup when a loose object vanishes mid-copy (#440's race, through a file copy) #645, auto maintenance is off in every repository the suite makes, so this test would guard a layer the fixture already covers.

Verify: the suites that hold the sites (tests/test_family_bump_remove_rollback.py, tests/test_doctor_legs_members_next.py, tests/test_doctor_gitlink_repairs.py, tests/test_clone_just_pushed.py) and tests/test_bump_leg_paste_lines.py plus the other read_through_git users, on Linux. A test of the helper: the prefix list for a typed and a resolved directory, with a trailing separator and without (tests/test_clone_just_pushed.py is the file for it).

Overlaps. Open PRs: none of the seven open PRs (#621, #627, #628, #630, #634, #639, #656) touches these files. Lane 5's in-flight branches: fix/645-fixture-copy-maintenance-race edits tests/conftest.py and adds tests/test_no_auto_maintenance.py (distinct hunks, :85 and :137, but the same file). Lane 4 holds none of these lines.

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

🤖 Generated with Claude Code