A citation the reader cannot follow is decoration, not provenance. Three defects stood between the archive’s sources and the people consuming them.
Fixed
- Saint
citationswere always empty. The reader queriedtabula_scopus='sancti'while both writers storeddevotio.sancti— the strings never matched, so every citation in the archive fell through the gap in silence. - Documented saint lookup routes were inverted and returned 422/404 when followed. The correct forms are
GET /v1/devotio/saints/slug/{slug},GET /v1/devotio/saints/{saint_id}, andGET /v1/devotio/saints/slug/{slug}/feast. - The Summa article example in the docs showed Latin field names (
quaestio_id,numerus,titulus). Responses use English (question_id,number,title).
Added
citations[].bibliography— volume, author, article and rights for cited print works. Previously written to the database and never served.citations[].source— resolves an internal citation toopus_slugandchapter_slug, so a saint’s life links to the very chapter it was drawn from:GET /v1/veritas/opera/{opus_slug}/read/{chapter_slug}.scripture_citations[].book_slugand.book_nameon Summa articles. Proof-texts previously carried only an internalbook_idthat consumers had no means of resolving.- Full schema tables for
CitationRead,CitationSourceRefandSummaScriptureCitationin the engine documentation.
Changed
- Documentation now states plainly that provenance is recorded only for hand-scribed saints. An empty
citationsarray is not a claim that no source exists — the backfill is deliberately human-reviewed, since name similarity alone cannot establish that an article is the life of a given saint.