Skip to content

docs(compat): close WP71-02 by measuring 7.1's design tokens - #175

Open
dknauss wants to merge 1 commit into
mainfrom
compat/wp71-02-design-tokens-measured
Open

docs(compat): close WP71-02 by measuring 7.1's design tokens#175
dknauss wants to merge 1 commit into
mainfrom
compat/wp71-02-design-tokens-measured

Conversation

@dknauss

@dknauss dknauss commented Aug 21, 2026

Copy link
Copy Markdown
Owner

WP71-02 asked for a decision, not a fix, on adopting WordPress 7.1's --wpds-*
design tokens. This measures it rather than arguing it, and closes it.

Risk: low. Planning docs only — no plugin code, no assets, no tests touched.

How it was measured

wp-env on 7.1.1-alpha-63326, driven with Playwright through the existing e2e auth
harness, reading computed --wpds-* values off :root and computed backgrounds off
#adminmenuwrap across five admin colour schemes. Contrast figures are derived from
those measured values with the WCAG 2.x relative-luminance formula.

The probe was a throwaway spec and is not committed — the register says so, and says
it should land env-gated like the capture specs if these figures ever become
load-bearing again.

What it found

Zero of 167 tokens track the admin colour scheme, while the surface Maestro draws on
does (five distinct menu backgrounds). Scheme-following was the entire prize; it is not
on offer.

The palette is Gutenberg's, not classic wp-admin's#fcfcfc vs #fff, #1e1e1e
vs #1d2327, #3858e9 vs #2271b1. Adopting would align Maestro with the editor by a
margin too small to read as deliberate.

So Phase 23's conclusion survives; only its wording had expired. The decision is now
recorded with a revisit condition rather than left stale.

The result worth reading: core's own --wpds-color-stroke-focus (#3858e9) fails
WCAG 1.4.11 against the menu surface on every scheme — 2.83:1 on fresh down to
1.46:1 on ocean/sunrise/coffee — where Maestro's existing #72aee6 passes all five
(6.74 / 4.78 / 3.48). Adopting core's token there would have been an accessibility
regression, which inverts the assumption that core's token is the safer default.

WP71-03 half closes on the same run

Phase 25's 6.74:1 and 9.11:1 reproduce exactly — and the entry now records why
they were never at risk: both were measured against Maestro's own hardcoded #1d2327
(maestro.css:536), not core's sidebar, so #65382 could not have moved them. That is a
correction to how the entry was framed when opened.

Adds five-scheme figures for chrome on the real menu surface, which had never been
measured on any version — all pass. Leaves open: core's own focus-state changes
(#65645, #65765, #65726), untested here; and a new hue question on the hardcoded
#3c434a divider against scheme-aware backgrounds.

Also in here

The file had outlived two of its own claims — #159 and #157 are described as open and
both have merged. Corrected to name their commits.

npm run check:doc-links clean. WP71-02's static counts re-derived unchanged on
572f472 (99 hex, 0 var(--)), with the command recorded inline.

🤖 Generated with Claude Code

WP71-02 asked for a decision on adopting 7.1's --wpds-* tokens. Measured it
instead of arguing it: wp-env on 7.1.1-alpha-63326, driven through the existing
e2e auth harness, reading computed values off :root across five admin colour
schemes.

Zero of 167 tokens track the admin colour scheme, while the menu surface Maestro
draws on does. The palette is also Gutenberg's rather than classic wp-admin's,
so every value is a near-miss (#fcfcfc vs #fff, #1e1e1e vs #1d2327, #3858e9 vs
#2271b1). Staying hardcoded keeps its conclusion and gains a live reason.

WP71-03 half closes on the same run. Phase 25's 6.74:1 and 9.11:1 reproduce
exactly, because they were always measured against Maestro's own #1d2327 rather
than core's sidebar — a correction to how that entry was framed. Adds the
five-scheme figures for chrome on the real menu surface, which had never been
measured. Core's own --wpds-color-stroke-focus turns out to fail 1.4.11 on every
scheme (2.83:1 down to 1.46:1) where Maestro's #72aee6 passes all five, so
adopting it there is contraindicated rather than safe.

Also corrects two claims the file outlived: #159 and #157 are merged, not open.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant