Skip to content

TESTING.md

Latest commit

 

History

History
68 lines (52 loc) · 3.43 KB

TESTING.md

File metadata and controls

68 lines (52 loc) · 3.43 KB
Raw

Delivera - Testing Guide

This project uses Playwright for end-to-end and integration tests. The root orchestration entrypoint is npm run test:all.

CSS pipeline before Playwright (SSOT)

  • npm run test:all always runs npm run build:css and npm run check:css before browser tests.
  • If public/styles.css is missing or out of sync with public/css/*.css, the run stops before Playwright starts.
  • npm run test:current-sprint:dedupe-fold also runs npm run check:css first for a faster Current Sprint gate.
  • CSS ownership/source-of-truth details live in public/css/README.md (do not duplicate them in this file).

Fail-fast gate order (after CSS)

npm run test:all runs these Playwright gates in order before the wider journey bundles:

  1. npm run test:journey:value-retention — journey.value-retention bucket (desktop density, alignment SSOT, investment drawer, period lens)
  2. npm run test:journey:direct-value-masterplan — cross-surface direct value contracts
  3. npm run test:focused — specs tagged @focused
  4. npm run test:journey:layout-overlap — governance/report/sprint overlap + mobile clip
  5. npm run test:current-sprint:dedupe-fold — sprint header/viewport declutter gate

Default behavior

  • npm run test:all is the full regression entrypoint.
  • The runner is foreground, serial, and fail-fast.
  • Each Playwright command is normalized to:
    • --max-failures=1
    • --workers=1
    • --reporter=list
  • The runner prints:
    • the active step name,
    • the exact command,
    • the primary spec/journey contract,
    • a periodic heartbeat while long steps run.

Optional impacted-only mode

  • Use IMPACTED_ONLY=1 npm run test:all for changed-only debugging.
  • In impacted-only mode, the runner:
    • checks changed files against TEST_BASE_REF (default origin/main fallback chain),
    • uses the saved last-failed spec list,
    • selects only impacted or last-failed steps,
    • still stays fail-fast.

Last-failed behavior

  • In impacted-only mode, failing specs are persisted to scripts/Delivera-Test-Last-Failed.json.
  • The next impacted-only run prioritizes those specs and applies Playwright --last-failed --pass-with-no-tests.
  • Disable that optimization with DISABLE_LAST_FAILED=1 npm run test:all.

Environment variables

  • IMPACTED_ONLY=1 - run changed-only / last-failed selection instead of full regression.
  • DISABLE_LAST_FAILED=1 - disable implicit last-failed optimization in impacted-only mode.
  • TEST_BASE_REF=<ref> - override the Git base ref used for impacted selection.
  • SKIP_NPM_INSTALL=true - skip the initial npm install step in orchestration.

New fail-fast UX trust validations

  • Added specs:
    • tests/Delivera-Report-Refresh-Trust-And-Action-Hierarchy-Validation-Tests.spec.js
    • tests/Delivera-CurrentSprint-Standup-Action-Rhythm-Validation-Tests.spec.js
    • tests/Delivera-Leadership-Mobile-FirstViewport-Decision-Validation-Tests.spec.js
    • tests/Delivera-CrossSurface-Context-Freshness-SSOT-Validation-Tests.spec.js
    • tests/Delivera-Duplicate-UI-Decision-Strip-Regression-Validation-Tests.spec.js
  • All new specs use the shared console/pageerror guard and runtime UI assertions (not brittle static copy checks).
  • Registering a spec in scripts/Delivera-Tests-Journey-Buckets-Map-SSOT.js is required for orchestration inclusion.
  • Expected network-abort simulation tests should use allow-console-pattern annotations when intentionally triggering ERR_FAILED.