tools/import/legacy-history-reference-source.test.ts esta en rojo y nadie lo estaba corriendo: el pin de cardinalidad espera 1 scanner crudo y hay 2
Medido por audit-mba-l-ca el 2026-09-23 al correr por primera vez el arbol de tests que ningun script ejecutaba: pnpm exec tsx --test 'src/**/*.test.ts' 'src/**/*.test.tsx' 'tools/import/**/*.test.ts' da 47 tests, 45 pasan, 2 fallan. Uno es vehiculos.line-transfer.test.ts (asercion vencida de expediente, va en MOVILBA-557). EL OTRO ES ESTE Y NO TIENE DUENIO. El pin espera UN solo scanner crudo y encuentra DOS. Eso no es ruido de test: o aparecio un segundo lector del crudo que nadie censo, o el pin quedo vencido cuando se agrego a proposito. Los dos casos importan, porque 'quien lee el crudo' es exactamente la distincion de MOVILBA-536 (proyectar lee el normalizado, materializar lee el CRUDO) y un segundo lector no contado es la forma de defecto que ahi costo dos vueltas. Primero se identifica cual es el segundo scanner y desde que commit existe; recien despues se decide si se corrige el codigo o el pin. NO se sube el numero del pin para poner el test en verde: un pin que se ajusta al hallazgo deja de ser un control. Queda FUERA del alcance de 557 a proposito (acordado con audit): 557 cablea src/**/*.test.{ts,tsx} via un script test:source y deja tools/import afuera para no arrastrar este rojo preexistente adentro de un ticket que no es suyo. Consecuencia: tools/import/**/*.test.ts sigue sin ejecutarse por ningun script hasta que este item se cierre - o sea sigue siendo un gate AUSENTE, no uno rojo.
Activity
-
coder04-mba-l-ca
-
coder03assistant-mba-l-ca
-
Reassigned per 2026-10-08 queue; expanding test:source to execute the import guard
-
2026-10-08: confirmed the historical raw-scan mismatch was corrected in 20fcf2ce (MOVILBA-558), which replaced the JS/TS hand scanner with TypeScript comment ranges and changed the live pin from 1 to 0. Current test asserts coverage.raw=0, source has no working-tree edits, and 20fcf2ce is reachable from dev, main, and origin refs. Gap remains: package test:source collected src/ only and test:tools collected tools/tests/*.test.*js, so tools/import/legacy-history-reference-source.test.ts was not executed by either ship gate. Updated tools/run-source-tests.mjs to collect tools/import tests alongside src tests. No tests run by this lane; audit should run the scoped test gate on this exact tree.
-
MOVILBA-558/515 audit (tree current as of 2026-10-08): SHA-256 of tools/run-source-tests.mjs = a614c36e...ab96, matches cited. Confirmed it now walks tools/import/**/*.test.ts in addition to src. Confirmed commit 20fcf2ce (MOVILBA-558, 2026-09-23) is an ancestor of both dev and main, and its subject documents the TS-parser comment-masking fix for exactly the rutap-historico.ts regex-vs-string case. Ran pnpm test:source on the current shared tree (src + tools/import): 64 tests, 63 pass, 1 FAIL - tools/import/movilba-518-projector-source.test.ts:68 asserts /d\.estado_id, d\.fecha_vigencia_desde, d\.fecha_vigencia_hasta/ against tools/rutap-snapshot-import.py. That literal no longer exists: commit 7f4c6c1a (MOVILBA-723, 2026-10-06, v5.6.179) legitimately changed the parity query to read fecha_vigencia_desde/hasta from the canonical v_vehiculos_export_vdc alias e instead of stored alias d (line 1299: 'd.estado_id, e.fecha_vigencia_desde, e.fecha_vigencia_hasta'), per that commit's own message. The test file's last commit (5aa5e65c) predates 723 and was never updated - this is a STALE test assertion, not a code regression. It was latent/non-blocking while run-source-tests.mjs only walked src; now that it also walks tools/import (the MOVILBA-558 change under review), it fails test:source on the current tree. Separately ran full pnpm test:tools for MOVILBA-515/714: 113/113 pass, 0 fail (includes the colectivo-edit-load.test.cjs dropdown fixture fix coder04 just shipped). Ran pnpm typecheck: clean. tools/tests/colectivo-expediente.test.cjs was left untouched per coder03assistant's note; its current content includes another lane's uncommitted MOVILBA-546/349/637 edits, not reviewed here. None of the 'four historical MOVILBA-515 failures' (incl. save-rejection) reproduce on this tree - test:tools is all-green.
-
Audit found that adding tools/import to the source gate surfaces one stale test: MOVILBA-518's source assertion still expects , but MOVILBA-723 commit 7f4c6c1a correctly reads dates from canonical export alias e and state from d. I am updating only that assertion under its exact-path lock, then will return to audit. Audit also measured MOVILBA-515: test:tools 113/113 and typecheck clean; none of the four historical failures reproduce.
-
coder03 11:19 ART: the runner change exposes a stale test: tools/import/movilba-518-projector-source.test.ts asserts /d\.estado_id, d\.fecha_vigencia_desde, d\.fecha_vigencia_hasta/ against the importer source and fails. Uncommitted change in the shared tree blocks every deploy (test:source is on both ship paths). Fix the 518 test or revert the runner until green.
-
Audit found that adding tools/import to the source gate surfaces one stale test: the MOVILBA-518 source assertion still expects d.estado_id plus d.fecha_vigencia dates, but MOVILBA-723 commit 7f4c6c1a correctly reads dates from canonical export alias e and state from d. Updated only the assertion to d.estado_id, e.fecha_vigencia_desde/hasta. Audit also measured MOVILBA-515: test:tools 113/113 and typecheck clean; none of the four historical failures reproduce.
-
audit pass: test:source on corrected tree, 2 files, SHA-verified
payload
{"scope":"test:source on corrected tree, 2 files, SHA-verified","outcome":"pass","evidence":"run-source-tests.mjs SHA a614c36e...ab96, movilba-518-projector-source.test.ts SHA fa3788bf...819f both match cited; pnpm test:source 64/64 pass, 0 fail","reportRef":null} -
coder03 11:26 ART: deployed v5.6.184 dev / v5.6.185 prod.
-
2026-10-08 deployment note: commit 0b47b800 contains both audited files and is on dev v5.6.184; coder03 reports prod v5.6.185. No pod version is recorded here, so I am leaving the item open for fleet-parity evidence.
-
coder03 11:35 ART: fleet parity v5.6.185 on PROD, DEV and pod (pod /api/health patch 185 via proxy-mba-w-cc).
-
coder03 11:35 ART: shipped v5.6.184/185, pod at 185, parity complete.