[Paullo9]

Requested by Paul · project thread

Fixes #40, together with #41 (this PR does the parts #41 left open). The two branches merge cleanly into each other and into main.

Fixture move: the nine review regression CIF headers (own commit)

The first comment line of alpha_quartz_P1_origin_shifted.cif, magnesium_P1_origin_shifted.cif, polonium_P1_supercell_7x.cif, mixed_centring.cif, rocksalt_P1_origin_shifted.cif, iron_P1_general_position.cif, corundum_P1_rhombohedral.cif, gold_P1_origin_shifted_exact_decimals.cif and magnesium_P63mmc_origin_shifted_loop.cif named a past review session. Each header now says which structure the file holds and what RefuterRegressionTests checks with it (for the 7x supercell, also the conventional-cell pins). Only comment lines change. Tests/Fixtures/README.md has the new sizes.

The pins do not change. These nine CIFs are not sha256-locked anywhere. Neither pymatgen-cells-pins.json nor numpy-structure-factor-pins.json lists them: their cif_file and cif_sha256 entries cover the other CIF fixtures only. To check this, I reran tools/pins/cells/pins.py, tools/pins/structure-factors/pins.py and tools/pins/conventional-cell/pins.py (the only generator that reads one of the nine, the polonium supercell) on the edited files, using the pinned environment (pymatgen 2026.9.24, pymatgen-core 2026.10.16, spglib 2.8.0, NumPy 2.5.4):

• numpy-structure-factor-pins.json: byte-identical.
• pymatgen-cells-pins.json and spglib-conventional-cell-pins.json: byte-identical except the generated_at stamp.
• Every self-check passes (6960, 7138 and 192).

No pin file is committed, so the pinned values and the sha256 values stay as they are.

Pin script comments

The comments, docstrings, READMEs and failure messages in tools/pins (cells/tetragonal-pins.py, structure-factors/pins.py and its README, conventional-cell/pins.py, phase-vectors/pins.py and its README, ring-calibration/ring-pins.py and q-pins.py, ring-indexing/pins.py) and in tools/consumer-check said slice 1, 2, 3a, 3b or 4. They now name the pin file, generator or rule they mean. Variables SLICE1_PINS, slice1, SLICE2_PINS and slice2 are renamed CELLS_PINS, cells_pins, STRUCTURE_FACTOR_PINS and sf_pins. The first docstring lines of the seven scripts that #41 edits are left to #41.

Not changed: strings written into fixtures

These strings are written into fixtures and stay as they are, because no fixture is regenerated here:

• the "source": "slice-1" label in the conventional-cell pins (so the ConventionalSymmetryTests filter stays too, as docs: name what tests, parity rows and pin scripts hold #41 documents)
• "purpose", "families" and "rings"
• the vs_slice2_* keys
• the predict labels

Change them the next time those pins are regenerated for another reason.

CHANGELOG

release-please hides test: entries by default. release-please-config.json now shows them under "Fixtures and tests", as in ImagoKit. The override below gives the fixture move its CHANGELOG line.

BEGIN_COMMIT_OVERRIDE
test: the nine review regression CIF headers say which structure each holds and what it tests (comment lines only; no pin or value changes)
END_COMMIT_OVERRIDE

Tests

• swift test -c release -j 1 --skip DiffractoKitParityTests --filter DiffractoCoreTests: 170 tests in 28 suites passed. This covers RefuterRegressionTests and the tests that read every CIF in the fixtures folder. CI runs the full unit gate.
• All edited Python scripts compile. The three generators above rerun with byte-identical output (apart from the timestamp). phase-vectors/pins.py was not rerun: it needs the py4DSTEM environment. Its changes are comments, failure messages and the two renamed variables, checked by search.

🤖 Generated with Claude Code