v0.1.14: Reproducible, Controllable Synthetic Data
- Added: --seed makes a run reproducible: the same seed gives the same data byte-for-byte, so generated fixtures hold up in tests and CI
- Added: --distribution shapes a column's values (zipf, normal) and --null-rate sets how often a column comes back null
- Added: --since and --until confine generated dates and timestamps to a window you choose
- Added: generated people stay coherent row by row, with name, email, birth date, age, and gender that agree
- Added: weavori sync --subset copies a referentially complete slice (a row count or a percentage), and --anonymize scrubs PII with stable, repeatable maps
- Changed: a formula or dataset aimed at a foreign-key column is now rejected up front, before any rows are generated, instead of silently breaking referential integrity