Skip to content

spec(wtc-rebuild): fold 2026-08-05 owner decisions - #718

Merged
WilfordGrimley merged 2 commits into
masterfrom
spec-wtc-rebuild-amend
Aug 6, 2026
Merged

spec(wtc-rebuild): fold 2026-08-05 owner decisions#718
WilfordGrimley merged 2 commits into
masterfrom
spec-wtc-rebuild-amend

Conversation

@WilfordGrimley

@WilfordGrimley WilfordGrimley commented Aug 5, 2026

Copy link
Copy Markdown

Summary

Records seven amendments (A1–A7) in the WTC rebuild spec. The spec is a living ground-truth document rather than a historical proposal: it shipped via PR #446 one day after it was written, and its container-first structure and tokens are still live in QuestionFeed.tsx.

  • A1 — container-first is binding, not proposed. The policy already governs the shipped implementation, so framing it as a proposal misdescribes the code.
  • A2 — the reference image is pinned within its container, not the viewport. WD4 rejects viewport-coupled positioning; snip container pinning satisfies persistent visibility without reintroducing it, and is expressible in the container-first idiom A1 states.
  • A3 — the §1c binding table stands unchanged. The reported bottom-corner obstruction is hover-scale clipping (What's That: hover-zoom on card/illustration is clipped by its container frame #705), not the gradient overlay the table prescribes, so no amendment to it is warranted.
  • A4 — the attribute chip panel is shown according to the type of question being asked rather than sitting behind a manual toggle, and must never occlude the reference image.
  • A5 — the Annex A served-mix logging seam is retired as a build item. Its purpose was correlating engagement against question difficulty; snip A6 derives difficulty from stored vote state instead, so the logging buys nothing the data does not already carry.
  • A6 — difficulty lanes derive from data state: confirm an uncontested machine vote, break a tie between competing votes, or answer cold with no prior. Lanes ramp per session with a user override. Deriving from vote state rather than observed behaviour is what makes A5's retirement possible.
  • A7 — records current implementation state and points at the epic tracking open work, so the document describes the surface as it actually stands.

Amendments follow the spec's existing amendment convention, mirroring the closure block at the document's top.

Test plan

…ruth document

Amendments A1-A7 record today's owner decisions:
- A1: reaffirm container-first policy as binding
- A2: reference image pinned within container (not viewport), reconciling WD4
- A3: confirm binding table; #708 symptom is hover clipping (#705), not gradient
- A4: AttributeChipPanel shown by question type, never occludes reference image
- A5: retire served-mix logging seam (A6 model requires no telemetry)
- A6: record difficulty-lane model (easy/medium/hard from vote state)
- A7: current implementation state and epic #704 children tracking

Spec is ground truth; amendments fit its own existing voice and amendment
convention (mirroring the 2026-07-24 Owner Questions section at top).
…was decided

A rationale that dissolves when its citation is removed was never a rationale,
so each amendment now carries the reason it holds:

- A1: container-first is binding rather than proposed because it already
  governs shipped code
- A2: the reference image must stay visible because it is the subject of the
  question; viewport-sticky remains rejected because it couples component
  layout to page-level scroll state
- A3: the reported obstruction is hover-scale clipping, not the gradient, so
  the binding table needs no change
- A4: a panel that must be opened before it helps is one most people never
  open, and the question already determines which chips are relevant
- A5: A6 derives difficulty from stored vote state, so the logging seam would
  record nothing the data does not already carry
@WilfordGrimley
WilfordGrimley merged commit a8a1e13 into master Aug 6, 2026
13 checks passed
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