Adopt CardRow and a SelectWrapper in one member
The pilot executor. Three members answer the open decisions with evidence; everything mechanical is inherited from the Button loop.
- Path
- loops/Adopt-CardRow-And-A-SelectWrapper-In-One-Member.md
- Authors
- Michael Staton
- Augmented with
- Claude Code on Claude Opus 5 (1M context)
- Tags
- Loop · Augment-It · Design-System · CardRow · Layout
Adopt CardRow and a SelectWrapper in one member
This is a PILOT, not a rollout
Three members, chosen to disagree with each other. Your job is as much to produce evidence as to produce a diff. Several decisions in [[../decisions/The-List-Surface-Paradigm-CardRow-ListContainer-And-Selection]] are deliberately open, and the pilots are how they get answered.
If the API does not fit your member, that is a RESULT, not a failure. Say so loudly and precisely. Do not contort the member to fit, and do not reach for rung 4 to paper over a bad API — an escape hatch used on the first try means the API is wrong.
Inherited, not repeated
Everything mechanical comes from
[[Adopt-The-Shared-Button-In-One-Member]] — the probe recipe, resolve.alias
(never source.alias) with the mandatory sentinel grep, measure-the-before,
gate capture, raise-don’t-chase, and the ladder. Read it in full.
Rung 4 is style=, not class= — corrected today. A member class is
(0,1,0) and loses to the component’s (0,2,0), while rendering perfectly and
looking like it worked.
What you are adopting
CardRow— one object in the list. A<div>, no onclick, not a control.SelectWrapper--ClickBody— clicking anywhere selects. Needs a positioned ancestor (CardRow is one) and console-errors if a sibling control would be buried under its overlay.SelectWrapper--ClickPrimary— only the primary label selects. No overlay.
The open decisions you are evidence for
| question | what your member should report | |
|---|---|---|
| D1 | one component or many? | did the thin base fit, or did you need a CardRow--<Kind> wrapper? If you wrote one, paste it — three identical wrappers means promote |
| D2 | naming | did any name need explaining to you? Name the one that did |
| D3 | which SelectWrapper | which variant, and why the other one was wrong here |
| D6 | is the pilot set right? | what your member has that the other two do not |
Rules
- Touch ONLY your member’s
src/**. Neverpackages/— if the primitive is wrong, say so; do not fix it. Three members editingshared-uiin parallel is how a component gets three incompatible APIs in one evening. - Do NOT commit. Report; the lead commits.
- Raise, don’t chase.
- Delete your probe before the final gate.
- Additive where you can. There is a demo on Tuesday. A member that ends the night half-migrated is worse than one not started — if you run short, finish fewer surfaces completely and say which you left.
Gates
node scripts/design-drift.mjs --member <prefix> # read the prefix from DESIGN.md
node scripts/design-drift.mjs # 71 now; moves for reasons not yours
pnpm --filter <pkg> check # 0 errors
pnpm --filter <pkg> build
Deliver
Gate output verbatim · before/after counts · which SelectWrapper and why · visual coverage stated explicitly (how many rows rendered and measured, and how) · an a11y delta led by target size and the overlay’s sibling-reachability, then contrast, then aria · findings raised not chased · and answers to the four questions above, which are the point.
Related
- [[../decisions/The-List-Surface-Paradigm-CardRow-ListContainer-And-Selection]]
- [[Adopt-The-Shared-Button-In-One-Member]] · [[Adopt-The-Shared-Chip-In-One-Member]]
- [[../specs/Component-API-Contract-And-The-Control-Scale]]