test(hrr): harden 2 vacuous-pass cache-persistence assertions (#707)

[robotrocketscience]

Closes #707.

What

Hardens two HRR cache-persistence tests in tests/test_hrr_struct_index.py
that pass vacuously when persistence is silently disabled:

• test_cache_loads_from_disk_on_second_construct
• test_cache_byte_equality_persist_round_trip

Both tests previously asserted only np.testing.assert_array_equal /
probe(...) equality between cold and warm caches. Under the in-memory
rebuild path (deterministic seed=7), those equalities hold even when no
disk round-trip happened. A future regression that silently disables
persistence would not be caught.

Hardening shape

Mirrors prior art in the same file (test_cache_persists_to_disk_after_build,
test_cache_load_uses_mmap):

1. After the cold construct, assert (persist_dir / "struct.npy").is_file()
   — proves the cold path actually wrote to disk.
2. After the warm construct, assert isinstance(b.struct, np.memmap) —
   proves the warm path loaded from disk via mmap rather than rebuilding
   in-memory.

Two atomic commits

1. test: harden vacuous-cache persistence assertions (#707) — the two
   strict assertions on each test.
2. test(hrr): isolate #707 hardened tests from ephemeral-path auto-disable
   — adds monkeypatch.setattr(hi, "_EPHEMERAL_PATH_PREFIXES", frozenset())
   to both tests, mirroring PR feat(hrr): ephemeral-path auto-disable for HRR persistence (#695) #701's pattern for the other 4 persist
   tests in this file. Required because tmp_path on Linux CI resolves
   under /tmp/, which after PR feat(hrr): ephemeral-path auto-disable for HRR persistence (#695) #701 triggers persist auto-disable —
   the new strict assertions would then fail loudly in CI without this
   neutralisation.

Verification

• uv run pytest tests/test_hrr_struct_index.py -q → 49 passed, 4
  skipped (Linux-only ephemeral-path tests skipped on macOS).
• Acceptance property verified locally: temporarily injecting
  monkeypatch.setenv("AELFRICE_HRR_PERSIST", "0") into each hardened
  test makes it fail at the new (pd / "struct.npy").is_file()
  assertion with the expected message. Reverted before commit.
• Both commits SSH-signed (G).
• Discretion grep on full diff vs github/main → clean.

Out of scope

• Refactoring the persistence substrate (issue scope).
• macOS /private/tmp quirk docs — slated for docs/CONFIG.md as
  part of the [v2.1] HRR integration: persistence default-ON + split-format migration + disk-cost opt-out #553 docs bundle (issue docs(hrr): CONFIG.md + COMMANDS.md + CHANGELOG bundle (#553 sub-task) #699).

Summary by CodeRabbit

• Tests
  • Enhanced struct index cache persistence tests with stronger validation to verify that cached data is properly persisted to disk and subsequently loaded efficiently for continued access.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 6b4b8ef8-c5e6-4d33-ab43-6cde4b1cfbdb

📥 Commits

Reviewing files that changed from the base of the PR and between 4480b85 and 3c7f587.

📒 Files selected for processing (1)

• tests/test_hrr_struct_index.py

---

📝 Walkthrough

Walkthrough

This PR strengthens two HRRStructIndexCache persistence tests by adding explicit verification that the persistence round-trip succeeds: struct.npy exists on disk after cold build, and the warm-loaded struct is mmap-backed.

Changes

HRRStructIndexCache Persistence Verification

Layer / File(s)	Summary
Disk persistence and mmap-backed loading assertions tests/test_hrr_struct_index.py	Two persistence tests add post-condition assertions: after cold build, struct.npy exists in the persist directory; on warm load, the loaded struct is np.memmap, explicitly verifying the persistence round-trip occurred.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~8 minutes

Possibly related issues

• #707: Changes the same two HRRStructIndexCache persistence tests with identical struct.npy existence and np.memmap assertions.
• #691: Tests directly exercise the HRRStructIndexCache persist/load/save/mmap wiring that the new assertions validate.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title directly and concisely summarizes the main change: hardening vacuous-pass cache-persistence assertions in HRR tests, and references the related issue #707.
Description check	✅ Passed	The description provides a detailed summary of the changes, explains the 'why', covers type of change (test-only), includes verification results, test plan, and notes for reviewers. Closely follows the template structure.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch test/issue-707-harden-vacuous-cache-persistence-asserts

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[sourcery-ai]

Sorry @robotrocketscience, you have reached your weekly rate limit of 2500000 diff characters.

Please try again later or upgrade to continue using Sourcery

[robotrocketscience]

[claim:review:Faraday:2026-05-12T02:08:40Z]

[robotrocketscience]

Reviewed by Faraday.

Rebased onto current main (2 commits ahead, was 2 behind). Both commits SSH-signed (G).

Verification

• Diff: 22 lines added to tests/test_hrr_struct_index.py, two assertions per test (cold-build struct.npy file exists; warm-load is mmap-backed). Matches the prior-art pattern at test_cache_persists_to_disk_after_build / test_cache_load_uses_mmap.
• uv run pytest tests/test_hrr_struct_index.py -q: 49 passed, 4 skipped (Linux-only ephemeral-path tests skipped on macOS, expected).
• _EPHEMERAL_PATH_PREFIXES = frozenset() monkeypatch correctly neutralises PR feat(hrr): ephemeral-path auto-disable for HRR persistence (#695) #701's auto-disable for these tests (they run under tmp_path which on Linux CI is /tmp/-prefixed).
• Discretion grep: clean.
• Acceptance property from PR body matches what the diff actually asserts.

Closes #707. Adding ready-to-merge.

[robotrocketscience]

[claim:review:pascal:2026-05-12T02:12:43Z]

[robotrocketscience]

[release:review:pascal:2026-05-12T02:12:48Z]

[github-actions]

merge-train: blocked

FF push to main failed:\n\n\nremote: error: GH006: Protected branch update failed for refs/heads/main. remote: remote: - All comments must be resolved. To https://github.com/robotrocketscience/aelfrice ! [remote rejected] 3c7f587b13560acbe7c25bb83edb7c980fb13e20 -> main (protected branch hook declined) error: failed to push some refs to 'https://github.com/robotrocketscience/aelfrice'\n\n\nCommon causes: branch protection rule changed, force-push detected by another writer, or token permission insufficient. Re-add the label after investigating.

The ready-to-merge label has been removed. Address the issue above and re-add the label when you're ready for another attempt.

[github-actions]

merge-train: merged 3c7f587 → main via FF push.

[robotrocketscience]

[release:review:Faraday:2026-05-12T02:16:23Z]