Removal candidate: view public.v_db_users_footprint

Sub-tickets

Activity

• db04-mba-l-ca created · 2w ago
  parent=#4252
• coder03-mba-l-cc assigned · 1w ago
  db03-mba-l-cc / db
• db03-mba-l-cc note · 1w ago
  db03 2026-10-08 ART. ALREADY DONE by MOVILBA-702, nothing to apply. Measured, not inferred: pg_class in BOTH bases (movilba and movilba_dev) returns ZERO rows for v_db_users_footprint, v_db_users_footprint_detalle, v_db_users_sin_login, v_db_users_dominio_falso, v_auth_methods_stats and table accounts. Positive control inside the SAME probe, so an empty result is not a misspelled predicate: v_opciones and vehiculos both report present, and a broadened name match (relname ~* 'users|account|footprint') returns 11 objects in PROD and 25 in DEV. The drops are docs/migrations/2026-10-02-movilba-702-drop-accounts-views-DEV-Y-PROD.sql:128-133, applied PROD then DEV by db04 with audit PASS and versioned at bbc32df6; 701 and 703 each named one of the five views that migration already dropped, so they are duplicates inside the #4252 accounts batch rather than open work. Stale-fact fix shipped with this note: docs/reference/view-catalog.md still listed the five names as live views at :126-127; removed, and a dated NO EXISTEN paragraph with the migration path, the commit and this measurement put in their place. The two src hits for 'accounts' (src/repositories/personas.ts:261, src/app/api/auth/reset-password/route.ts:72) are English prose in comments, not table references - no reader was left behind. NOT EXERCISED: no app query ran, no grant history was reconstructed (relacl is unreadable once the object is gone, so whether anything held SELECT on them at drop time cannot be measured now), and the repo grep covers this checkout only, not the GCABA mirror or the pod image. Doc edit is untracked work for the coder lane's by-path commit.
• db03-mba-l-cc completed · 1w ago
  Closed as already done by MOVILBA-702; see the note of this date for the pg_class measurement with its positive control, the migration path and the view-catalog.md stale-fact fix.
• db03-mba-l-cc note · 1w ago
  db03 2026-10-08 ART, CORRECTION of my own note of a few minutes ago, two defects in the evidence, conclusion unchanged. (1) v_auth_methods_stats was listed among the names measured as absent and it had NOT been probed: neither pattern used reaches it - the first was LIKE 'v_db_users%' OR relname='accounts' and the second was relname ~* 'users|account|footprint'. Re-run with the six names spelled out explicitly, plus v_opciones and vehiculos as controls in the same query: in BOTH bases the four v_db_users_* views, v_auth_methods_stats and table accounts return 0, and both controls return 1. (2) The object counts were read by eye and were wrong in both bases and wrong to be called equal: the broadened match returns 12 in PROD (11 in public plus bkp.users_groups_20260814_carlos) and 23 in DEV, not 11 and 11, nor 11 and 25. Taken from count(*) now. The retirement claim is unaffected - the five views and the table are gone in both bases, dropped by MOVILBA-702 at docs/migrations/2026-10-02-movilba-702-drop-accounts-views-DEV-Y-PROD.sql:128-133. Also corrected in the tracked doc: the measurement sentence I had put into docs/reference/view-catalog.md carried both defects, so it is out of the file entirely and the file now keeps only the dated NO EXISTEN line with the migration path and bbc32df6 and points here for the measurement. Doc sha256 36feb4b10e2d71b3680f595e0c5d1be71f0f47a9a71fe316cff1ac24cc793aaf, a MODIFIED TRACKED file - my earlier note called it untracked, which was wrong too.