NEWS

vahtian 0.2.0

Per-field epistemic states on assertion(): a plain value is explicitly stated and NULL is not stated, while inferred(), ambiguous(), not_applicable(), and extraction_failed() record the other reasons a field holds (or lacks) a value — so a single NULL no longer collapses distinct states. vahtian_compare() surfaces the most actionable absent state per field (extraction_failed > ambiguous > not_applicable > not_stated) and records each side's state in the assessment; an inferred key field never reaches an "aligned" candidate without a human, and an extraction failure is kept distinct from source silence. Comparator format bumped to vahtian-compare/2; the cross-language golden hash is updated in lockstep.

assertion() / vahtian_compare() / assessment_record(): reduce a claim and a cited source's finding to the same structured assertion, compare them field by field (deterministic, no model), and record the run in the audit ledger. Assertion and assessment hashes are byte-identical with the Python package, covered by the cross-language golden-hash test. Checks claim–source support, not truth.

First release.
vahtian_freeze() / vahtian_verify() / vahtian_content_hash(): freeze a record set into a content-hashed, provenance-stamped, date-locked corpus and verify it (tamper-evident).
ledger() / ledger_append() / ledger_verify(): a hash-chained, tamper-evident audit trail.
The canonical format and content hash are byte-identical with the Python package 'vahtian', verified by a cross-language golden-hash test.