Files
memento-frontend/tests/visual/MEMENTO_VISUAL_BASELINE.md

73 lines
3.1 KiB
Markdown
Raw Permalink Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
# Memento visual baseline provenance
## Authority
The visual authority is the approved Beta prototype captured from Git commit
`85c804b0301091b36fc48a83ae501570340269e7`, preserved under
`tests/fixtures/memento-visual-baseline/legacy/`.
The normal 26-surface Playwright snapshots are regression evidence for the React
implementation. The separate legacy-parity suite proves that those React
surfaces remain within the approved prototype rendering envelope; snapshot
files are not treated as self-authorizing design truth.
## Rendering conditions
- Browser: repository Playwright Chromium
- Desktop viewport: 1600 × 1100
- Mobile viewport: 390 × 844
- Auth viewport: 1440 × 1000
- Color scheme: light
- Locale: en-US
- Media: identical `public/assets` bytes for both renderers
- Map iframe: hidden in both renderers before capture
- Animations: disabled before capture
- Fonts: same host/browser font environment for both renderers
The comparison requires identical screenshot dimensions. Pixel differences are
measured from two screenshots produced in the same Playwright run.
## Difference policy
A visual diff is not fixed by updating a snapshot.
1. Identify the DOM/CSS source of the diff.
2. Confirm that layout dimensions, product colors, spacing, and typography
values still come from the approved design contract.
3. If the change is intentional, review and approve the design change first.
4. Only after approval may the React regression snapshot be updated.
The legacy-parity gate uses two independent limits. First, any changed pixel
must remain at or below `0.5%` of the surface. Second, a material pixel—one
whose maximum RGBA channel delta exceeds `13/255` (roughly 5%)—must remain at
or below `0.2%`. This separates harmless font/native-control rasterization from
meaningful layout or color drift. Screenshot dimension mismatch always fails.
A diff above either envelope must be investigated; increasing either limit is
not an acceptable way to make the test green.
## Accessibility interaction with the visual lock
The approved legacy palette contains known WCAG AA color-contrast debt. During
this migration the visual authority has higher precedence than an unreviewed
palette change: changing those colors would invalidate the direct legacy ↔
React parity evidence.
For routes owned by `memento-web`, the generic Axe E2E suite therefore waives
**only** the `color-contrast` rule. Critical/serious findings from every other
Axe rule remain blocking. This is an explicit migration debt, not an assertion
that the current palette is accessible.
Removing this waiver requires a separately approved Memento design-system
change. That change must update the visual authority/baselines and accessibility
evidence together; the waiver must not be broadened to other Axe rules.
## Commands
- React regression: `corepack pnpm test:visual`
- Approved legacy ↔ React: `corepack pnpm test:visual:memento-legacy`
- Snapshot update after explicit design approval only:
`corepack pnpm test:visual:update`
The legacy fixture itself is immutable migration evidence. Regenerating it from
the current React implementation is prohibited.