Repository navigation
v0.7.2
Movement caches now identify the supplied data and analysis settings, preventing a repeated run from reusing an unrelated dated result.
- Identical requests reuse the exact CSV without fitting or rewriting it. Changed data, filters or settings produce a separate
movement-<entity>-<cadence>-<16-character fingerprint>.csv. use_cache=Falserefreshes only the matching file atomically. Failed fitting or saving preserves the previous result; unreadable keyed files raise a clear error.- Inactivity-window and prior settings are forwarded through the wrapper. Text identifiers, numeric precision and Monthly CSV compatibility are preserved.
The first call after upgrading may recompute: legacy dated caches are left untouched and are not reused. No dependency upgrades or changes to analytical rules or AutoViz are included.
Validation: 164 maintained tests, Ruff, strict documentation, locked dependencies and package checks pass. PR CI passes all nine Linux/macOS/Windows × Python 3.12–3.14 combinations, including installed-wheel exports and offline reopening.
What's Changed
- Fix movement cache reuse across data and settings by @okkymabruri in #47
Full Changelog: v0.7.1...v0.7.2