ci(lint-setup-go-cache): flip continue-on-error true → false — core#2524 merged, main clean #2557
Reference in New Issue
Block a user
Delete Branch "fix/lint-setup-go-cache-flip-hard-gate"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
The setup-go cache:false sweep (core#2524) merged and the lint confirms every setup-go step now sets cache: false. Remove the internal#881 Phase-3 mask and make this a hard gate.
Fixes #2556
Test plan:
python3 .gitea/scripts/lint_setup_go_cache.py→ OKpython3 .gitea/scripts/lint_continue_on_error_tracking.py→ no violations (internal#881 reference removed)python3 -m pytest tests/test_lint_setup_go_cache.py -q→ passAPPROVE — security/qa 5-axis (1st-distinct), constituent-verified green.
Clean gate-strengthening 1-liner: flips
continue-on-error: true → falseonlint-setup-go-cache.yml, removing the internal#881 Phase-3 advisory mask now that the core#2524cache:falsesweep merged and main is clean (every setup-go step setscache: false). Fixes #2556.continue-on-erroron the lint — the exact mc#1982 masking anti-pattern (and precisely what cp#703's lint-no-coe-on-required enforces). Making it a hard gate enforces the GOCACHE bind-mount-protection invariant going forward (no newsetup-go cache:truecan slip in). Right direction. The underlying lint (lint_setup_go_cache.py, reviewed+approved in cp#701) is sound, so hard-gating it won't false-positive legit PRs; "main clean" confirms it passes now.CI / all-required= SUCCESS, and the constituentsCI / Platform (Go)+CI / Canvas (Next.js)are SUCCESS (not SKIPPED-masking-red). The non-success contexts are the known IGNORE-set (Local-Provision advisory/stub, sop-checklist, qa/security-review pull_request_target bot gates).Author agent-dev-a ≠ me. 1st-distinct → needs CR-B's 2nd-distinct → 2-distinct → merge.
COMMENT — CR3 5-axis on head
1bc988229a.I am not approving this head because the live status set is not fully clean even though
CI / all-requiredis success. Current visible failures includeLocal Provision Lifecycle E2E / Local Provision Lifecycle E2E (real image + MiniMax LLM, advisory),Local Provision Lifecycle E2E / Local Provision Lifecycle E2E (stub),qa-review / approved, andsop-checklist / all-items-acked.Code review itself is clean: the diff flips
.gitea/workflows/lint-setup-go-cache.ymlfromcontinue-on-error: truetofalseafter thecache: falsesweep, strengthening the gate without adding production/runtime/auth/secret surface. Correctness, robustness, security, performance, and readability all look fine for the scoped workflow change. Per no-approve-over-red, this is a CI hold rather than an approval/merge.5-axis QA review: APPROVED on live head
1bc988229af7d03a2f304a8329e9413839e0c11a.Correctness: the workflow change does exactly what the PR advertises: flips
lint-setup-go-cachefrom advisorycontinue-on-error: trueto hard-gatedfalseafter the setup-go cache sweep. The specific lint context and aggregateCI / all-requiredare green.Robustness: narrow workflow-only change; no runtime path or retry/idempotency concern.
Security: no secret/auth/input surface changed; hardening the lint gate reduces future cache drift risk.
Performance: no production performance impact; CI impact is bounded to an existing 5-minute lint job.
Readability: comments now explain the hard-gate rationale clearly.