Run the bats suite on main pushes (coverage parity with unit) — re-scoped after #1485 correction #1486

Activity

1. don-petry

   added

   enhancementFeature requestsFeature requests

   initiativeEpic / initiative tracking issueEpic / initiative tracking issue

   on Aug 8, 2026
2. don-petry

   added a parent issue on Aug 8, 2026

   • Formalize the agentic PR-interaction model (event-first, timers-as-backstop) — clear, repeatable, enforceable #1402
3. don-petry

   changed the title [-]Run unit/bats suites on main pushes — trunk merged corrupted twice with zero post-merge signal[/-] [+]Run the bats suite on main pushes (coverage parity with unit) — re-scoped after #1485 correction[/+] on Aug 8, 2026
4. don-petry

   added

   dev-leadFor dev-lead agent pickupFor dev-lead agent pickup

   on Aug 18, 2026
5. don-petry commented on Aug 18, 2026

   

   don-petry

   on Aug 18, 2026

   CollaboratorAuthor

   More actions

   Dev-Lead: issue #1486 implementation failed — will retry

   The engine failed (engine-error — e.g. a timeout or transient engine error). This was attempt 1 of 3; dev-lead will retry automatically. You can also re-apply the dev-lead label to retry now.

   • Cause: engine-error
   • Run: https://github.com/petry-projects/.github-private/actions/runs/32156056431

   Last 40 lines of engine output (redacted)

       [copilot] calling gh copilot (model=openai/o4-mini, timeout=2700s, flags=--yolo)
   Error: Classic Personal Access Tokens (ghp_) are not supported by Copilot.
   
   The GH_TOKEN environment variable contains a classic PAT.
   Please use a Fine-Grained Personal Access Token or another authentication method.
   
   To fix this, you can:
     • Replace the token in GH_TOKEN with a fine-grained PAT
     • Unset GH_TOKEN and run 'gh auth login' to authenticate
     • Unset GH_TOKEN and start 'copilot', then use the '/login' command
   
   
6. don-petry commented on Aug 18, 2026

   

   don-petry

   on Aug 18, 2026

   CollaboratorAuthor

   More actions

   Dev-Lead Implementation Plan

   Issue: #1486 — Run the bats suite on main pushes (coverage parity with unit) — re-scoped after #1485 correction

   Scope

   Add a paths-filtered push: branches: [main] trigger to lint.yml so the bats job re-runs against trunk state after every merge, and exclude the known-red template-drift job from main-push runs (#1448) — no change to PR-triggered behavior.

   Implementation Checklist

   • Add push: branches: [main] with paths scripts/**, tests/**, .github/workflows/** to the on: block of .github/workflows/lint.yml (AC1)
   • Add if: github.event_name != 'push' to the template-drift job so it stays excluded from main-push wiring until [bug] seed-repo-template fetches standards unpinned — pin to published stable, compliance accepts N-1 (continues #1436) #1448 is resolved (AC4)
   • Leave the existing pull_request + workflow_dispatch triggers untouched — no new required PR checks (AC2)
   • Register the new guard test in lint.yml's bats list

   Tests to Write

   • tests/test_lint_main_push_trigger.bats — assert lint.yml has push: branches:[main] with the three paths, that the pull_request trigger is unchanged, and that template-drift carries the event_name != 'push' guard

   Test Command

   scripts/run-bats.sh tests/test_lint_main_push_trigger.bats (full suite: the bats job's file list in lint.yml)

   AC3 note

   Fleet Monitor already tracks lint.yml failures across all events/branches (fleet_monitor.sh queries actions/workflows/<id>/runs with no event filter, and lint.yml is not in FLEET_GATE_WORKFLOWS), so a failing main-push run surfaces in its reports with no extra wiring — matching #1484. No on-failure issue-open step needed.
7. don-petry commented on Aug 18, 2026

   

   don-petry

   on Aug 18, 2026

   CollaboratorAuthor

   More actions

   Dev-Lead: Implementation Complete

   Plan Execution

   • Added paths-filtered push: branches: [main] trigger to .github/workflows/lint.yml on: block, scoped to scripts/**, tests/**, .github/workflows/** (AC1)
   • Added if: github.event_name != 'push' to the template-drift job so it stays excluded from main-push wiring until [bug] seed-repo-template fetches standards unpinned — pin to published stable, compliance accepts N-1 (continues #1436) #1448 is resolved (AC4)
   • Left the pull_request + workflow_dispatch triggers and PR paths untouched — no new required PR checks (AC2)
   • Wrote tests/test_lint_main_push_trigger.bats guard and registered it in the bats job's file list
   • Confirmed AC3 needs no wiring (see Notes)

   Test Results

   1..5
   ok 1 lint.yml declares a push trigger scoped to the main branch (AC1)
   ok 2 lint.yml main-push trigger is paths-filtered to the bats-relevant paths (AC1)
   ok 3 lint.yml still triggers on pull_request — PR behavior unchanged (AC2)
   ok 4 lint.yml pull_request paths are unchanged by the push wiring (AC2)
   ok 5 template-drift is excluded from main-push runs until #1448 is resolved (AC4)
   

   Repo lint (.dev-lead/scripts/dev-lead-lint.sh): all checks passed (shellcheck on 110 files + agent-profile/CODEOWNERS validation).
   scripts/validate-workflow-schedules.sh: OK — no minute-0 crons introduced.
   yamllint (ci.yml config) on lint.yml: passes (the sole comment-indentation warning at the caller-stub-freeze block pre-exists on main and is untouched).

   Files Changed

   • .github/workflows/lint.yml: added the main-push trigger block, added the event_name != 'push' guard to template-drift, and registered the new bats file. Trigger/job-guard changes only — no script or job restructuring ([Phase 4] Name and guard the "agent optimizes the check, not the intent" failure mode #1468).
   • tests/test_lint_main_push_trigger.bats (new): pins the wiring so a future template sync cannot silently drop the main-push trigger (AC1), alter PR-trigger behavior (AC2), or wire template-drift into the push path before [bug] seed-repo-template fetches standards unpinned — pin to published stable, compliance accepts N-1 (continues #1436) #1448 is resolved (AC4).

   Notes

   • AC3 (Fleet Monitor visibility) — verified, no code needed. scripts/fleet_monitor.sh queries repos/<repo>/actions/workflows/<id>/runs with no event/branch filter, so a failing lint.yml run on a main push is counted like any other failure; lint.yml is not in FLEET_GATE_WORKFLOWS, so a genuine failure surfaces in the report / high-failure tracking. This matches what petry-projects — workflow failures detected 2026-08-08 #1484 suggested, so no on-failure issue-open step was added.
   • Whole-workflow push, per the issue's chosen option. Adding the trigger to lint.yml's on: runs every non-excluded lint job on main pushes (bats included). AC4's explicit template-drift carve-out confirms this whole-workflow approach was the intended path; only the known-red template-drift ([bug] seed-repo-template fetches standards unpinned — pin to published stable, compliance accepts N-1 (continues #1436) #1448) is excluded.
   • First post-merge run should start green (trunk restored by fix: restore trunk scripts to single-definition state — surgical dedup of engine.sh + 3 more (#1485) #1487).
8. don-petry

   added a commit that references this issue on Aug 18, 2026

   feat: implement issue #1486 — Run the bats suite on main pushes (cove…

   ...

   18e7c6d
9. don-petry

   closed this as completedin #1530on Aug 18, 2026