Repository navigation
docs(rfc): record the foldable layout's Phase 0 probe readings and test table - #33463
jeeshofone wants to merge 1 commit into
Conversation
First Principles Review (Opus 5.5, fork) — ⏭️ skippedRevision |
Design Review (Opus 5.5, fork) — ✅ PASSDesign-level review of Design-Verdict: PASS A docs-only change to a draft RFC: it records measured readings, and the one new precedence rule waits for a named maintainer's sign-off. Notes
[DESIGN-REVIEWED] ba38298 Description read[DESCRIPTION-READ] f43cd38e2df46bd3c0418e2266a1633e9e6948d76eb5f0626a754822c0b87aa6 That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
GPT 6.1 Review (fork) — ✅ no blocking findingsReviewed Review detailsNo findings. |
Opus 5.5 Review (fork) — ✅ no blocking findingsReviewed |
dec640a to
7dd8a08
Compare
UX Review (Opus 5.5, fork) — ⏭️ skippedRevision |
7dd8a08 to
d3d0d0b
Compare
d3d0d0b to
b57cac4
Compare
Code change approver recommendation: A person should decideWhy: the decider estimates a 10% chance that this change introduces a defect, within the 12% limit for approving on its own, but it is held at tier 2: design-doc-gate: D. Docs-only update to a draft RFC that records Phase 0 simulator probe readings and a test table, answers Open question 1, links the fix for the virtualizer bug, and proposes one precedence rule (a reported fold beats the #30619 touch-first phone rule at ≥640px), with maintainer sign-off explicitly requested. Verified at head: nothing under website/src reads the fold APIs, so the "Nothing built" index row is correct, and the touch-first rule the new section cites exists as described (touchLayoutForm.ts, the md: variant keyed on data-layout-form, narrow-viewport.md). The status stays draft, split view is marked unmeasured rather than passed, and CI and all lanes are green. No code, state or trust boundary changes. Tier: 2, held for a person: design-doc-gate: D. An automated recommendation, not a review: a maintainer still decides. It approves on its own only when the estimated chance of a defect is at most 12% and the change has at most 400 changed lines. Decider v3c (217fe607) · commit |
|
Re the code change approver note (comment): the description now matches the diff. It no longer says the diff only adds rows and rewords one sentence, or that the signed design is unchanged. It names everything the commit changes beyond the two tables:
The precedence rule itself still needs a maintainer to sign it. |
f4ab380 to
edab496
Compare
edab496 to
293dbe3
Compare
293dbe3 to
d0d38f9
Compare
|
Rebased onto current main (89438fd) to clear a conflict in the docs/request-for-change/README.md index row; the RFC text is unchanged. |
Problem / Motivation
Goal: Put the foldable layout RFC's Phase 0 probe readings and test table into the document, so Phase 1 can start from measured fold signals.
The fold-aware layout RFC (
docs/request-for-change/rfc-foldable-viewport-segments.md) says no Phase 1 code opens until its Phase 0 table is in the document, and the maintainer's sign-off on #29145 repeats that. The table was not there.Why it matters
Phase 1 builds fold detection on what the browser reports. Without measured readings, that code would rest on guesses about which poses report a fold.
Not a goal
What changed (motivation → approach → change)
The RFC's Phase 0 section now holds two tables. The first records what the iPhone Duo simulator (Safari 27.2, both feature flags on) reports in six poses, in a Safari tab and as a Home Screen web app: posture, both segments media queries,
window.viewport.segmentsand everyviewport-segment-*env()value. The second gives one row per pose and window width with its expected layout, the fold signal behind it and how it is tested. That covers outer and inner displays, half-folded, split view, Home Screen web app, windows 639px to 768px wide with and without a fold, Reduce Motion and right-to-left.The readings answer Open question 1: the two half-folded segments meet with no gap. A fully open inner display in portrait (669px) reports one segment, and half-folded in portrait reports two stacked segments, so under the signed rule both keep the phone form. In the web app the fold sits at 475.5px of an 867px viewport, so fold placement has to read the segments, not the viewport centre.
The "That bug is on main today" sentence in Continuity now links #29523, which fixed it. The front matter and the index row record the re-audit at
87247b9cdb.The document also gains one design rule, under Design, "A reported fold and the touch-first phone rule": when the fold query matches two horizontal segments and the viewport is at least 640px wide, the two-pane form takes precedence over the touch-first phone rule from #30619, which left fold-aware layout to this RFC. A flat inner display at 951px, the inner display upright at 669px and every touch browser that reports no fold keep the phone form. The test table rows for those poses now say so.
@bolichen97, you signed the direction of #29145: please sign that precedence line, since it meets a merged rule. The RFC's status is unchanged by this PR; on
mainits front matter still readsstatus: draftand its first line reads "- Status: draft. Nothing is built."Backwards compatibility
Compatible: documentation only.
Removes nothing from the design. Besides the two tables, the diff changes these lines:
last-audited,audited-at,doc-pr: 29145andtracking-issues: [29143].87247b9cdband replaces "Acceptance requested from a maintainer; Phase 0 (a probe) runs first" with the Phase 0 status. The RFC staysdraft, so acceptance is still pending.Tests
Proof: not provable -- docs or config only
./scripts/docs-lint.shandsemgrep scan --config semgrep/both pass on this head.Manual verification
The readings come from a probe page on the iPhone Duo simulator that uses the dashboard's own viewport meta tag, opened in a Safari tab and from the Home Screen. Each pose was set with the simulator's fold and rotate controls with both flags on; a flags-off pass shows no API in any pose. Split view could not be entered (Device Hub's menus have no split control and the App Switcher shows one card per app), so it has no reading and no test result.
Related Issues
Part of #29143
Checklist
feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)Re-dated for a fresh CI run after the repository sweep cancelled two re-runs of the previous run; content unchanged (tree identical).