AI578HM y las 15 grafias del mismo chasis Agrale GNC: alias en chasis_resolver, no catalogo nuevo
Cierra la pregunta 93 del tracker SIN mandarsela a Eugenia. AI578HM, fila_excel 5549 del libro vigente 2026-09-30-1305, trae Chasis_Modelo = '52-MT 17.0 GNC CA 7'. Medido sobre las 5548 filas con modelo del libro. VEREDICTO: es un error de tipeo, no un hueco de catalogo. La vinculacion a AGRALE MT 17.0 LE GNV (opcion 123) es CORRECTA y no hay que tocar el catalogo. Evidencia, toda del propio libro: 1. El numero de chasis de AI578HM es 8BBC84B1AVM000086. El bloque 8BBC84B agrupa 89 filas, 87 con Unidad_Propulsora GNC, todas Piso Bajo, motor WEICHAI y carroceria TODO BUS: es UN solo chasis fisico. Sus vecinos inmediatos de la misma serie AVM -AI501JV 000074, AI482RD 000075, AI513QP 000081, AI563JM 000082- escriben 'MT 17.0 GNV CA 6200'. 2. Ese mismo bloque de VIN aparece escrito de QUINCE maneras distintas: MT 17, MT 17.0, MT 17.0 GNC, MT 17.0 GNV, MT 17.0 GNV CA, MT 17.0 GNV CA 6200, MT 17.0 GNV V 6200, MT 17.0 LE, MT 17.0 LE 6000, 52-MT 17.0 GNV, 52-MT 17.0 GNV CA 62, 52-MT 17.0 GNV CA620, 52-MT 17.0 GNV CA6200, 52-MT 17.0 GNC CA 7 y la larga 'MT 17.0 GNV CA 6200 7.8 6 272 AUTR NEU'. La dispersion es de tipeo, no de modelo. 3. El prefijo '52-' no es parte de ninguna designacion: aparece en 5 filas de este mismo chasis, y la misma forma se repite con '20-' y '45-' en otros modelos Agrale. Es un numero de fila pegado adentro de la celda. 4. GNC por GNV es el uso argentino del combustible en lugar de la designacion del fabricante; el libro usa los dos indistintamente sobre el MISMO VIN. 5. 'CA 7' es la cadena larga cortada. La unica fila que la trae entera, 'CA 6200 7.8 6 272', descompone en distancia entre ejes, cilindrada, cilindros y potencia. La investigacion online confirma la cilindrada y la potencia del MT 17.0 GNV: 7.8 litros, seis cilindros en linea, 272 CV a 2200 rpm (camionargentino.blogspot.com/2019/06/agrale-mt-170-gnv.html, camionesybuses.com.ar). El 6200 de distancia entre ejes NO se corroboro: Agrale publica 6000 y 6500 para otras variantes, asi que ese numero queda como dato del libro, no verificado contra el fabricante. QUE HACER: cargar las 15 grafias como alias de la terna (AGRALE, MT 17.0 LE GNV, GNC) en rutap.chasis_resolver, para que la proxima variante de tipeo resuelva sola en vez de abrir otra pregunta. Medir primero cuantas filas resuelven hoy y cuantas no. HALLAZGO APARTE, mismo bloque de VIN: 2 de las 89 filas declaran Unidad_Propulsora GO en lugar de GNC -AI259HP fila 3123 es una, con Tipo_Motor E5-. La propulsion es parte de la clave del catalogo, asi que un GO ahi resuelve a otro chasis. Medir las dos y reportar; no se corrige sin decir cuales son.
Activity
-
db04-mba-l-ca 2026-10-01 11:35 ART. SELECT-only PROD movilba, snapshot 8 (2026-09-30-1305, 5548 source rows): VIN prefix 8BBC84B has 89 rows, 87 GNC and 2 GO. GNC rows: 85/87 currently resolve to chasis 123 (AGRALE MT 17.0 LE GNV), 2/87 to generic chasis 135 (MT 17), none NULL. All 15 distinct GNC spellings already have approved rutap.chasis_resolver keys: 13 point to 123 and the two generic strings MT 17 and MT 17.0 point to 135; there are no missing keys to INSERT. The 2 GO domains are AI259HP fila3123 (MT 17.0 GNV CA 6200, GO, currently chasis 123 by an older provisional GO resolver) and AI325HY fila5181 (MT 17, GO, chasis 135). This corrects the item's broad claim that GO always selects a different chassis: one GO row currently still selects 123. No GO mapping will be touched. Same current book is in DEV snapshot 20: 85/87 GNC history rows on 123, 2 on DEV generic chasis 143; target 123 is identical semantic row in both bases. Caution: rutap.proyectar_snapshot refuses re-projection once history exists, so alias edits alone change the resolver view but not the two already-stored history rows; measuring both layers after the edit is required. No writes yet.
-
db04-mba-l-ca 2026-10-01 11:37 ART. Alias-only SQL draft /data-lezama/movilba/migration-evidence/2026-10-01-movilba-635-agrale-gnc-alias-DEV-Y-PROD.sql, shared instrument. ROLLBACK rehearsals passed independently in PROD movilba snapshot8 and DEV movilba_dev snapshot20, same pinned 2026-09-30 book md5 8512833be84d1568743d61b601179c6b, 5548 rows. Each rehearsal: 2 existing GNC keys changed (MT 17, MT 17.0), 15/15 VIN-block GNC spellings -> target chasis123; resolver-view cohort 87/87 GNC target123 and GO 1/2 target123, while stored history remains 85/87 GNC target123 because projection refuses rerun. No persistent write yet; audit verdict requested. Correction to previous event: measurement time was 11:33 ART, not 11:35.
-
audit pass: Static review of 2026-10-01-movilba-635-agrale-gnc-alias-DEV-Y-PROD.sql: snapshot pin+latest guard, cohort/baseline preflight (89/87/2, 15/13/2), UPDATE predicate scoped to GNC MT17/MT17.0 only, postcondition 15/15, final SELECT exposes stored-history gap.
payload
{"scope":"Static review of 2026-10-01-movilba-635-agrale-gnc-alias-DEV-Y-PROD.sql: snapshot pin+latest guard, cohort/baseline preflight (89/87/2, 15/13/2), UPDATE predicate scoped to GNC MT17/MT17.0 only, postcondition 15/15, final SELECT exposes stored-history gap.","outcome":"pass","evidence":"Static SQL review only, no DB creds from this lane. Relies on db04's two independent ROLLBACK rehearsals (PROD snapshot8, DEV snapshot20). UPDATE touches only unidad_propulsora_id=157 rows at v_old_id; GO rows and catalog untouched by predicate. Alias-only confirmed, does not re-project stored history (85/87 stays).","reportRef":null} -
db04-mba-l-ca 2026-10-01 11:38 ART. Audit PASS recorded. Applied audited SHA256 03ad8252aceb16e71c042596ab06f6fc05495e400335084df2c510d35372dfd7 under rutap.chasis_resolver locks, PROD movilba first then DEV movilba_dev, same SQL. Both COMMITs succeeded; independent post-COMMIT SELECT: snapshot8 PROD and snapshot20 DEV each have 87/87 GNC VIN-block rows resolving via v_rutap_chasis_resuelto_aprobado to chasis123; both generic GNC keys MT 17 and MT 17.0 now approved ->123. Two GO keys untouched. Stored imported history remains 85/87 on 123 in each base; two rows retain generic chassis because the installed projector refuses re-projection. No history/catalog row was changed. Artifact is now docs/migrations/2026-10-01-movilba-635-agrale-gnc-alias-DEV-Y-PROD.sql, same SHA as audited/applied. Alias part is complete; current-history repair remains unassigned/undecided, so item is not closed.
-
db04-mba-l-ca 2026-10-01 11:42 ART. Post-alias read-only finding: the two lagging current-book imported rows are AI501KK fila4008 and AI563JN fila4448. PROD history ids127109/127507 and vehicles7007/7024 still hold generic chasis135; DEV history315818/315645 and vehicles6700/6754 still hold generic chasis143. Both bases' resolver now returns 123 for those exact AGRALE/model/GNC keys. Installed history UPDATE trigger protects raw_* and date fields, not chasis_id, so deliberate derived-FK repair needs no trigger disable; vehicle UPDATE will archive old image (+2 expected). DEV vehicles are ssot_import and no app audit_log UPDATE was found; audit_log cannot prove absence of direct SQL edits. Same follow-up SQL /data-lezama/movilba/migration-evidence/2026-10-01-movilba-635-two-current-history-rows-DEV-Y-PROD.sql: ROLLBACK rehearsals passed on final bytes in PROD and DEV; each updated exactly two history rows and two vehicles, detail view displayed MT 17.0 LE GNV, archive +2, then rolled back. GO exceptions from current book: AI259HP fila3123, Tipo_Motor E5, GO, provisional GO resolver maps to 123; AI325HY fila5181, Tipo_Motor GNC, GO, generic 135 in PROD. The provisional AGRALE/MT 17.0 GNV CA 6200/GO alias matches ONLY AI259HP in all retained rutap_rows in PROD and DEV (each stores current snapshot only). No GO key changed. Awaiting audit before persistent repair.
-
audit pass: Static review of 2026-10-01-movilba-635-two-current-history-rows-DEV-Y-PROD.sql: snapshot/resolver preconditions, STRICT single-row match per domain, DEV ssot_import+audit_log guard, archive +2 check, detail-view postcondition, no trigger disable, raw fields untouched.
payload
{"scope":"Static review of 2026-10-01-movilba-635-two-current-history-rows-DEV-Y-PROD.sql: snapshot/resolver preconditions, STRICT single-row match per domain, DEV ssot_import+audit_log guard, archive +2 check, detail-view postcondition, no trigger disable, raw fields untouched.","outcome":"pass","evidence":"Static SQL review only, no DB creds from this lane. Relies on db04's two independent ROLLBACK rehearsals post-alias-commit. UPDATE touches only the 2 named history+vehicle rows by id; GO rows AI259HP/AI325HY not in predicate. Minor: archive count check is by-domain not by-row, negligible given txn-local row lock.","reportRef":null} -
db04-mba-l-ca 2026-10-01 11:46 ART. Audit PASS on follow-up recorded. Applied SHA256 a53131c2c7acf06c2fe3ba35ee55c92007677e616e083d53bc7756956e28a8cf under exact table locks, PROD movilba first then DEV movilba_dev, same SQL, both COMMIT. Each leg: two imported history rows and two live vehicles chasis_id generic -> 123, archive trigger produced exactly two old-image rows, v_vehiculos_detalle displays MT 17.0 LE GNV for AI501KK and AI563JN. Independent post-COMMIT SELECT: current VIN-block GNC history AND resolver view both 87/87 target123 in PROD snapshot8 and DEV snapshot20. No raw source column or GO row/key changed; no trigger disabled. Applied artifact now docs/migrations/2026-10-01-movilba-635-two-current-history-rows-DEV-Y-PROD.sql, exact same SHA. Pending repository commit/push before closure; GO exceptions and provisional alias reach already recorded separately in this item.
-
Both audited SQL artifacts were applied PROD then DEV (2026-10-01 verified events), and committed as 12ed562e. Current source cohort reached 87/87 GNC VIN-block mappings to chassis 123 in resolver and stored history; two GO exceptions were reported without modification. No remaining action in this item's scoped alias/history repair.