EPS
← All tasks·#1382Completed

workflow-fix: fix two red workflow tests on main

kind: infra#wf-fix#wf-fix-fp:12839736f95a#daily-auto-filed
track:

Highlight text on any card (body, plan, or an event) to anchor a comment, or leave a whole-task comment here. Mention @claude to summon a reply.

No comments yet.

Overview / Motivation

Auto-filed by the /daily 2026-07-15 parked-candidate routing pass (Step C) from a recursion-guard-parked prose follow-up on task #1326 (emitting agent: implementer r1; both items discriminator-confirmed NOT introduced by #1326's diff).

Goal

Fix two PRE-EXISTING red workflow-invariant tests on main: (1) tests/test_living_docs.py::test_check_degrades_on_drifted_registry_path — stale vs find_task_path's registry-drift self-heal (task_workflow.py ~:3521); update the test to the self-heal contract. (2) tests/test_workflow_followup_labels.py::test_corpus_replay_all_historical_markers — #1092's 2026-07-11 followup-scope marker carries an unparseable label; fix the corpus-replay expectation or retro-repair the marker parse.

Workflow gap

  • Bug observed: two workflow-invariant tests fail on pristine main, poisoning every Step 9c test-verdict baseline that selects them.
  • Why it is a workflow gap: red-on-main workflow tests force every session into per-run pre-existing-failure discrimination and can mask genuine regressions.
  • Confidence (emitter): medium
  • verified-at-filing: uv run pytest tests/test_living_docs.py::test_check_degrades_on_drifted_registry_path tests/test_workflow_followup_labels.py::test_corpus_replay_all_historical_markers → BOTH FAIL on current main (AssertionError at test_living_docs.py:532 — REGISTRY stale for #207 approved→completed; AssertionError at test_workflow_followup_labels.py:1124), re-run 2026-07-16 UTC at filing time (test-run evidence; grep n/a for a behavioral red-test claim)

Proposed change (candidate diff sketch — refine in planning)

(1) update test_check_degrades_on_drifted_registry_path to the find_task_path self-heal contract; (2) fix the corpus-replay expectation for #1092's unparseable followup_label or retro-repair the marker parse.

Scope / surfaces

  • Primary target: tests/test_living_docs.py, tests/test_workflow_followup_labels.py
  • Note: the living-docs failure surfaces a REGISTRY drift for task #207 (tasks/approved/207 vs on-disk tasks/completed/207) — the planner should check whether task.py audit --repair --apply is part of the fix or a separate concern.

Constraints / invariants

  • Workflow-surface only — never experiment code, configs/, or tasks/ bodies (a task.py audit --repair registry re-sync via the canonical API is allowed if the plan calls for it).
  • Never weaken a test to make it pass — encode the CURRENT intended contract.
  • This session runs under EPM_WORKFLOW_FIX_SESSION=1 and carries a workflow_fix_target: Provenance line — it MUST NOT auto-route any of its own subagents' workflow-fix candidates (recursion guard).

Provenance

  • fingerprint: 12839736f95a

  • workflow_fix_target: tests/test_living_docs.py

parked prose follow-up (verbatim, from #1326 events.jsonl 2026-07-15T13:41:29Z): "(1) target_file: tests/test_living_docs.py — test_check_degrades_on_drifted_registry_path PRE-EXISTING red on main; stale vs find_task_path's :3521 registry-drift self-heal; proposed: update the test to the self-heal contract. (2) target_file: tests/test_workflow_followup_labels.py — test_corpus_replay_all_historical_markers PRE-EXISTING red on main; #1092's 2026-07-11 followup-scope marker carries an unparseable label; proposed: fix the corpus-replay expectation or retro-repair the marker parse. Both discriminator-confirmed NOT introduced by #1326's diff."

Activity