Skip to content

refactor(members): split MembersPage.tsx into focused owners - #20342

Merged
chenmingwei23 merged 1 commit into
mainfrom
refactor/linecap-f07-memberspage
Oct 15, 2026
Merged

chenmingwei23 merged 1 commit into
mainfrom
refactor/linecap-f07-memberspage

Conversation

@bolichen97

@bolichen97 bolichen97 commented Oct 14, 2026 •

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: Bring the Crewmates page's source file, website/src/pages/members/MembersPage.tsx, under the repository's 2,000-line cap by moving it into focused owner modules, with no change to what the page does.

MembersPage.tsx is 5,633 lines. One component, MembersPage, fills 4,725 of them. It is the third-largest hand-written frontend file. It changed 43 times in the last week, so every feature PR on the Crewmates page edits one giant function, and reviewers and agents read the whole thing to find the part they need.

Why it matters

The line-cap campaign (#20192) makes 2,000 lines a hard limit for hand-written files, and this file is nearly three times that. Splitting it by responsibility means the next change to the roster, the thread or the Profile card lands in a file about that thing. That file is at most about 1,050 lines, with room to grow before it nears the cap again.

Not a goal

  • No behaviour, layout, copy, test-id, i18n key, query key or route change. This is a structural move.
  • No rewrite of the moved code: bodies move verbatim, including their comments.
  • No split of MembersPage.test.tsx (scope FT06, after this lands) and no change to any assertion.
  • No new feature, hardening or clean-up beyond what the move itself needs.

What changed (motivation → approach → change)

Nothing on the Crewmates page changes for a user. Its 5,678-line file becomes a 924-line page plus thirteen owner files, each holding one part of the page, so the next change to the roster, the thread or the Profile card lands in a file about that thing.

The page was one component, and React runs its effects in the order its hooks are declared. A split that regrouped code by topic would reorder those registrations, which is a behaviour change even when every test still passes. So the split works in hook order. Each owner hook is one stretch of the page's hook sequence, and the facade calls it exactly where that stretch sat. Every useState, useRef, effect and library hook therefore registers in the same order as on main. A TypeScript-AST recipe generates the split from the base file and fails if that order differs.

File Owns
MembersPage.tsx (924 lines) the composition and the page's selection state; the New crewmate card and its follow-up; guide facts; exit guards; face-flight overlay; re-exports of every name it exported
membersPageModel.ts the pure rules: landing choice, memory reading, panel placement, withheld views, create-link parser, shared keys
useMembersPageRoster.ts, MembersPageRoster.tsx, MembersPageRosterRows.tsx roster data and live readings; roster view state and the roster column; one roster row and the empty roster's hero
useMembersPageCrewEditor.ts, useMembersPageTeams.ts, useMembersPageLayout.ts, useMembersPageDrawers.ts, useMembersPageNavigation.ts the in-place editor, teams, column placement, phone drawers, which crewmate is open
MembersPageThread.tsx, MembersPageThreadHeader.tsx, MembersPageProfile.tsx, MembersPageSidePanel.tsx the thread, its header, the Profile card and the side panel, each beside the state it renders

A few things stay in the facade because tests or lint read them there: the New crewmate guide wiring, the reveal scopes, the three allowlisted window.confirm guards and CrewFaceFlight. Owners never import the facade. The new MembersPage.owners.test.tsx pins the exports, the one-way imports and the order in which the facade calls its owners. One new behaviour test, MembersPage.unreadColumn.test.tsx, covers the thread header's two unread-column handlers, which nothing exercised; without it the moved MembersPageThreadHeader.tsx falls under the 80% per-file coverage floor every new file must clear (79.3%, 92% with it). Five source-pin tests follow the code: each positive check reads the file that renders what it pins, and each negative check reads every page file, so a hand-rolled copy cannot hide in another owner. Moved comments that said "above" or "below" about code now in another file name that file. The feature map and the crew-mode spec name the owners and map the Schedules draft guard.

Move audit

website/src/pages/members/MembersPage.tsx: 5678 -> 924 lines; moved 97.30% (line-multiset)
   unmatched removed lines: 132

The 132 removed lines that do not reappear verbatim are glue, a named comment pointer or a named state type:

Kind Lines Why
comment pointer named 46 moved comments that said "above", "below" or "this file" about code now in another file name that file and symbol instead (L29); an audit read every positional pointer in the moved comments against the base
imports 28 each file imports what it uses, from the same specifiers (so every vi.mock still applies)
a deps array gains a stable value 28 a BASE useState setter or useRef object now arrives through an owner object, so the hooks lint cannot see it is stable and asks for it; its identity never changes, so no memo recomputes and no effect re-runs
export added 15 moved declarations another owner imports
JSX that became a component 8 {placeRoster(<aside … </aside>)}, the Profile / side panel IIFE and the header IIFE became <MembersPageRosterColumn>, <MembersPageProfileCard>, <MembersPageSidePanel> and <MembersPageThreadHeader>
a named state type 7 the facade's profile, gone and postCreateError useState calls name their shapes (ProfileCardOpen, GoneNotice, PostCreateFailure in membersPageModel.ts), which the owners' option types share

No hook changed position: the recipe refuses to generate unless the 209 hook-registering statements run in the base's order, and every apply prints final order identical to BASE, so there is no moved-hook list (L40). For every changed file, git's default diff and --minimal add the same number of lines (L36).

Diff: 803,839 bytes (git diff BASE...HEAD). The recipe (a TypeScript-AST script, anchors not line numbers) regenerates this commit byte-for-byte from the base.

Backwards compatibility

Compatible: every import path, export, default export, test id, i18n key, query key and route stays the same. App.tsx's lazy route and the capture entries still import ./pages/members/MembersPage. Its named exports are the same bindings, now re-exported from membersPageModel.ts.

Removes nothing: no file, export, key or behaviour is removed; code moves to sibling modules the facade composes.

Re-points and citations (path or name only)

File Change
website/src/test/listShellParity.test.ts, website/src/test/kiroLightShellHooks.test.ts, website/src/test/searchFilterBar.parity.test.tsx, website/src/test/ResizeHandle.parity.test.tsx positive checks read MembersPageRoster.tsx (and, for the row recipes, MembersPageRosterRows.tsx), which hold the roster card and rows; negative checks read the facade and every owner through readMembersPage() (Master grant)
website/src/test/membersPageFamily.ts (new test helper) the one place the rule for "every file of the Crewmates page" lives; the owners test and the four parity pins read it, and it throws when it no longer finds the facade and the roster column
website/src/test/mobilePanels.compositor.test.ts reads MembersPage.tsx + useMembersPageDrawers.ts together (Master grant)
website/src/components/listShell.ts, website/src/components/Glass.tsx, website/src/apps/command-bar/matesProvider.ts, website/docs/theming-contract.md comments naming where the roster card, the identity pill and memberMemoryDisplay live
website/src/pages/members/MembersPage.test.tsx, website/src/pages/members/TeamDialog.test.tsx one comment each naming the file that now holds LAST_MEMBER_KEY and the "+" menu
docs/feature-map/README.md Crew Members row lists the owners; rows 111, 134-138, 244, 248 and 305 name the owner that holds their construct
.github/file-line-cap-baseline.txt deletes the MembersPage.tsx row: the file is now 909 lines, under the cap (only that row, by hand; scripts/check_file_line_cap.py reports 0 failures)
docs/system-specs/modules/crew-mode.md new owner table (where new Crewmates-page work goes) and a map of the Schedules draft guard across owners

CI path references

None: no workflow path filter or explicit file list names MembersPage.tsx.

Tests

  • New src/pages/members/MembersPage.owners.test.tsx pins:
    • the facade's named exports are the model's own bindings;
    • no owner imports the facade, and nothing outside the page imports an owner (specifiers resolved, @/ included, compared as paths);
    • each owner hook is called once, by the facade;
    • the owner calls and the page code the facade keeps appear in order.
      The order pin guards the split, not a design: a later change that regroups state by responsibility (follow-up refactor(members): move the Schedules exit-guard state into its owner hook #20512) reorders registrations deliberately and edits it.
      Mutations turn it red: an owner importing the facade (relative, and through @/), a module outside the page importing an owner, the shared file rule losing a family, and two owner calls swapped.
  • Re-pointed (Master grant), each proven by mutations (mutations.md in the campaign reports):
    • listShellParity, kiroLightShellHooks, searchFilterBar.parity and ResizeHandle.parity: positive checks read the roster files, negative checks read every page file through the shared src/test/membersPageFamily.ts. Removing the construct from its new home turns each red, and so does planting the forbidden markup in an owner other than the roster, or setting the helper's file rule to match nothing;
    • mobilePanels.compositor's "every gesture consumer declares its claim" check reads the whole page through the same helper, because the drawer gestures moved to useMembersPageDrawers.ts. With the old single-file read, a facade that dropped data-owns-swipe stayed green because the page was skipped.
    • Pins that walk the source tree read every owner file by construction; a violation planted in an owner turns safeArea.guard and englishIdentity red.
  • New src/pages/members/MembersPage.unreadColumn.test.tsx: picking another crewmate's unread pill under the switcher chip opens that crewmate, and "+N" opens the switcher's list instead. Nothing exercised these two handlers before. The moved header file needs them to clear the per-file 80% floor on its own (92% with them, 79.3% without). Passed 5 runs alone and in the 35-file parallel run. Mutating either handler to a no-op turns it red.
  • Unchanged and green: the 8 MembersPage.*.test.tsx files, MembersPanelVisibility, GuideLayer, revealScopes, safeArea.guard, uiIndex, autoGuide, the i18n ratchets, chatSlice.ownership, fileNameCasing, MemoI18nSubscriptionRatchet, App.mobileSingleTopbar and others: 35 files, 1,026 tests.
  • Per-file coverage (check_per_file_coverage.py, floor 80%): all 14 page files pass. Lowest: useMembersPageCrewEditor.ts 81.8%, MembersPageProfile.tsx 89.7%, MembersPageSidePanel.tsx 90.7%, MembersPageThreadHeader.tsx 92%.
  • Gates run locally:
    • tsc and eslint src --max-warnings 0;
    • i18n:check (0 findings on added lines, all key refs resolve);
    • lint:phantom-classes and lint:restyle-ratchet;
    • gen-ui-index --check (fresh) and jscpd;
    • analyze build with the bundle-size and chunk-cycle gates;
    • semgrep, diff-scoped: 0 new findings;
    • the brand, comment-history, focus-cue, feature-map and docs-lint gates.
  • local-gate.py --dry-run selects 821 backend and 383 frontend related files. That is too many to run locally (the floor rule), so the gate floor above ran instead; CI runs the full suites.

Manual verification

N/A — unit coverage sufficient: the change moves code verbatim, the page's behaviour is pinned by its existing 355 page tests, and the hook order is pinned by the new owners test.

Screenshots / video

Why no screenshot: code moves verbatim into sibling modules under the same element tree, so nothing rendered changes.

Related Issues

Part of #20192

Checklist

  • At most two commits (one is the norm), with a Conventional Commits title (feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)
  • Existing tests pass and new tests added for new functionality
  • Self-review completed; code follows project style guidelines
  • Documentation updated (if applicable)
  • No secrets, credentials, or internal references in the diff

Contribution License Agreement

Contributed under the project's Apache-2.0 license.

@bolichen97
bolichen97 requested review from a team and KiroCrewCaptain as code owners October 14, 2026 15:02
@bolichen97
bolichen97 requested a review from mbriones98 October 14, 2026 15:02
@github-actions github-actions Bot added the readiness: checking Automated validation is still running label Oct 14, 2026
@github-actions

github-actions Bot commented Oct 14, 2026 •

Copy link
Copy Markdown
Contributor

First Principles Review (Opus 5.5) — ✅ PASS

Premise-level review of 5c4a952ac59ac5279017f66b320f9a246265f32c — why this exists and whether the shipped surface is the smallest honest version. Updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

First-Principles-Verdict: PASS

One thing to check: the claim that hooks still run in the base order is proven by the author's out-of-tree recipe; the owners test only pins the order the facade calls its owners.

What this change ships

Inventory (7 items) — 7 justified

Intent: bring the Crewmates page's source file under the 2,000-line cap (an AGENTS.md invariant) with no change to what the page does. It is an ADDITION of structure only, not a fix.

  1. MembersPage.tsx goes from 5,678 to 924 lines, with the rest moved verbatim into thirteen owner files and every export path kept — justified
  2. The page's row in .github/file-line-cap-baseline.txt is removed, so the baseline shrinks — justified
  3. New MembersPage.owners.test.tsx pins the exports, the one-way imports and the order the facade calls its owners — justified
  4. New MembersPage.unreadColumn.test.tsx covers the thread header's unread handlers, which the per-file coverage floor requires — justified
  5. New test helper membersPageFamily.ts, read by 6 files (owners test plus 5 source pins) — justified
  6. Five source-pin tests re-point at the owner files; their negative checks now read every page file, and no assertion was removed — justified
  7. Comments and docs (feature map, crew-mode spec, theming contract, 3 source comments, 2 test comments) now name the new owner files — justified

[FIRST-PRINCIPLES-REVIEWED] 5c4a952

Description read

[DESCRIPTION-READ] dc8934b1cbb1e045e3042ed10c7ef4c23f50ee806be2201e03f23ba687781396

That sha256 names the PR description this verdict read. Recompute it with .github/scripts/pr-description-capture.sh; if it differs, the description changed after this verdict was formed and any finding drawn from the description is unproven.

@github-actions

github-actions Bot commented Oct 14, 2026 •

Copy link
Copy Markdown
Contributor

Design Review (Opus 5.5) — ✅ PASS

Design-level review of 5c4a952ac59ac5279017f66b320f9a246265f32c — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

Design-Verdict: PASS

Nothing to check.

[DESIGN-REVIEWED] 5c4a952

Description read

[DESCRIPTION-READ] 9bdc633943d992aa5b1a9217eda7c8719d0939d3238cab7669277ec7338c3327

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 .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions

github-actions Bot commented Oct 14, 2026 •

Copy link
Copy Markdown
Contributor

UX Review (Opus 5.5) — ✅ PASS

UX-level review of 5c4a952ac59ac5279017f66b320f9a246265f32c — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

UX-Verdict: PASS

Before merging, open the Crewmates page and check that the roster, thread header, Profile card and side panel look and behave exactly as on main.

[UX-REVIEWED] 5c4a952

Description read

[DESCRIPTION-READ] c2755360b0836e2a3e46455fb66a8dcce006605b4b7567dfcbdd4f3e55feee6f

That sha256 names everything this verdict read: the PR description, plus the 3 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 .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions

github-actions Bot commented Oct 14, 2026 •

Copy link
Copy Markdown
Contributor

Opus 5.5 Review — ✅ human override accepted

Reviewed 5c4a952ac59ac5279017f66b320f9a246265f32c — this comment is updated in place on each push.

Human judgment by @bolichen97 overrides the Opus 5.5 finding for 5c4a952ac59ac5279017f66b320f9a246265f32c; the recorded reason is authoritative for this commit.

[OPUS-OVERRIDE] 5c4a952

Verdict recorded from an authorized human decision for commit 5c4a952ac59ac5279017f66b320f9a246265f32c.

False positive or not applicable? A repository writer can comment:
/ai-review override fable 5c4a952ac59ac5279017f66b320f9a246265f32c: <one-sentence reason>

@github-actions github-actions Bot added the merge conflict Branch has merge conflicts with its base — author must resolve before merge label Oct 14, 2026
@github-actions

github-actions Bot commented Oct 14, 2026 •

Copy link
Copy Markdown
Contributor

GPT 6.1 Review — ✅ no blocking findings

GPT 6.1 completed its review of 5c4a952ac59ac5279017f66b320f9a246265f32c and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] 5c4a952

False positive or not applicable? A repository writer can comment:
/ai-review override gpt 5c4a952ac59ac5279017f66b320f9a246265f32c: <one-sentence reason>

@github-actions github-actions Bot added readiness: passed Eligible automated validation passed for the current revision and removed readiness: checking Automated validation is still running labels Oct 14, 2026
mbriones98
mbriones98 previously approved these changes Oct 14, 2026

@mbriones98 mbriones98 left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Checks green and PR Readiness passed at 3b72a1f; non-feature change.

@bolichen97
bolichen97 force-pushed the refactor/linecap-f07-memberspage branch from 3b72a1f to f7768ac Compare October 14, 2026 17:22
@github-actions github-actions Bot added readiness: checking Automated validation is still running and removed readiness: passed Eligible automated validation passed for the current revision merge conflict Branch has merge conflicts with its base — author must resolve before merge labels Oct 14, 2026
@bolichen97

bolichen97 commented Oct 14, 2026 •

Copy link
Copy Markdown
Contributor Author
  • span=a70e4d996d5f PAGE_HOOKS outlives the move (fixed in 1370095)

Fixed: the full-sequence pin is removed; the facade owner-order anchors remain (MembersPage.owners.test.tsx, "are called by the facade in the page's hook registration order").
The split's own hook order is proven by the recipe instead: split.mjs refuses to generate unless BASE's 206 hook-registering statements keep their order, and prints "final order identical to BASE" on every apply.
No list a later Crewmates PR must edit when it adds a hook.

@bolichen97

Copy link
Copy Markdown
Contributor Author
  • span=d0f97d3dab69 MembersPage.unreadColumn.test.tsx rides along (rebutted)

The move needs it: MembersPageThreadHeader.tsx, moved verbatim, measures 79.3% line coverage alone, under the 80% per-file floor every new file must clear (check_per_file_coverage.py); with the test it is 92%.
It covers two existing handlers nothing exercised (picking an unread pill, and "+N"), touches no production code, and the body's Tests section names it with that reason.
"No split of MembersPage.test.tsx" is about that existing file, which this PR leaves alone apart from one comment.

@bolichen97

Copy link
Copy Markdown
Contributor Author
  • The 28 deps-array additions do not change when effects re-run (answered)

Each added value is a base useState setter or useRef object that now reaches the code through an owner object, so the hooks lint can no longer prove it stable; the recipe lists all 28 (deps-added.txt) and the move audit counts them.
React keeps both identities for the component's lifetime and compares deps with Object.is, so none of these arrays differs between renders: no effect re-runs and no memo recomputes because of them.
The related suite (35 files, 999 tests, the MembersPage specs included) passes unchanged.

@bolichen97

Copy link
Copy Markdown
Contributor Author
  • Smoke test against main (answered)

Built main b6916a1 and this head and ran the gateway-free capture harnesses on both: capture-crewmate-card-rail-exit passes on both and its 10 frames (light and dark) are pixel-identical; capture-crewmate-panel-ia's 3 frames are pixel-identical too.
Their later steps fail identically on main (panel-ia's card opens on Sessions, so it times out on the Profile tab's read-more; unread-badge and patrol crash on a stale crew-board stub), so the harnesses are stale, not this split.
Behaviour: the MembersPage specs (375 tests) and the related suite (999 tests) pass unchanged, bodies moved verbatim, and the page's hook registration sequence equals main's.

@bolichen97
bolichen97 force-pushed the refactor/linecap-f07-memberspage branch from f7768ac to 1370095 Compare October 14, 2026 19:02
@bolichen97

Copy link
Copy Markdown
Contributor Author

self-added: yes

  • span=ce530ff38197 Give both chained thread-open waits explicit ceilings (fixed in 1370095)

Both waits now pass { timeout: THREAD_OPEN_TIMEOUT_MS } (5000 ms), and each carries a comment naming the chain it waits on: roster read, URL sync, thread-open POST, pane mount.
The constant's doc says why the 1 s default is too short for that chain on a loaded shard.
The test passed 5 runs alone after the change, and making either handler a no-op still turns it red (mutation proofs).

@bolichen97

Copy link
Copy Markdown
Contributor Author
  • span=ed53e0f62892 MembersPage.unreadColumn.test.tsx is undeclared (fixed in the description; the test stays)

The move needs the test: MembersPageThreadHeader.tsx, moved verbatim, measures 79.3% line coverage on its own, under the 80% per-file floor every new file must clear; with the test it is 92%.
It adds no production code and no hardening: it drives two existing handlers nothing exercised (picking an unread pill, and "+N").
The description's What changed section now names the file and that reason (it was only in the Tests section, past the part the lanes read).

@bolichen97

Copy link
Copy Markdown
Contributor Author
  • Is the new unread-column test described? (answered)

Yes: What changed now names MembersPage.unreadColumn.test.tsx and why it is here (the moved thread header needs it to clear the 80% per-file coverage floor: 79.3% without, 92% with), alongside the Tests section that already listed it.

@bolichen97

Copy link
Copy Markdown
Contributor Author
  • Open the Crewmates page and compare with main (answered)

Built main and this branch and ran the gateway-free Crewmates capture harnesses on both: capture-crewmate-card-rail-exit's 10 frames (light and dark) and capture-crewmate-panel-ia's first 3 frames are pixel-identical.
The harnesses' later steps fail the same way on main (stale fixtures), so they cannot show the rest; the MembersPage specs and the related suite pass unchanged, and every moved body is verbatim.
A maintainer's hand check before merge is welcome and is not something this PR can substitute.

@bolichen97

Copy link
Copy Markdown
Contributor Author

/ai-review override fable 1370095: Opus discovery failed closed on two runs (38620315506 attempts 1 and 2; no [OPUS-DISCOVERY] marker, 82-92 captured bytes, no denials), which is an infrastructure fault and not a finding; I verified the head myself — production code is identical to the independently reviewed split (range-diff vs 7e8618f1 changes only MembersPage.owners/unreadColumn tests and the commit message) and GPT, Design, First Principles and UX lanes pass.

@github-actions

Copy link
Copy Markdown
Contributor

Human judgment recorded

@bolichen97 marked the fable AI finding as false positive, not applicable, or explicitly accepted for d66ab1565d6bb19a80f8229f9d5ec1c2785bfdf1.

Opus discovery failed closed (run 38682245905, no [OPUS-DISCOVERY] marker), an infrastructure failure. Range-diff shows the patch equals f47e2fd008, whose #20476 port the Master verified against upstream.

This decision applies only to this commit. A new push requires a new judgment.

@github-actions github-actions Bot added merge conflict Branch has merge conflicts with its base — author must resolve before merge readiness: passed Eligible automated validation passed for the current revision and removed readiness: checking Automated validation is still running labels Oct 15, 2026
@bolichen97
bolichen97 force-pushed the refactor/linecap-f07-memberspage branch from d66ab15 to 45027a0 Compare October 15, 2026 05:48
@github-actions github-actions Bot added readiness: action required A blocking check or review needs attention merge conflict Branch has merge conflicts with its base — author must resolve before merge and removed merge conflict Branch has merge conflicts with its base — author must resolve before merge readiness: passed Eligible automated validation passed for the current revision labels Oct 15, 2026
MembersPage.tsx held one 4,725-line component in a 5,633-line file,
far past the 2,000-line cap the repo is adopting. The page now composes
owner modules beside it. Each owner is called where its code sat in the
page's hook order, so every useState, useRef, effect and library hook
registers in the same order as before.

- membersPageModel.ts: the page's pure rules and shared constants
- useMembersPageRoster.ts, MembersPageRoster.tsx,
  MembersPageRosterRows.tsx: roster data, view state, the roster column
  and its rows
- useMembersPageCrewEditor.ts, useMembersPageTeams.ts,
  useMembersPageLayout.ts, useMembersPageDrawers.ts,
  useMembersPageNavigation.ts: the in-place editor, teams, column
  placement, the phone drawers and which crewmate is open
- MembersPageThread.tsx, MembersPageThreadHeader.tsx,
  MembersPageProfile.tsx, MembersPageSidePanel.tsx: the thread, its
  header, the Profile card and the side panel, each with the state it
  renders

Bodies move verbatim. The facade keeps the New crewmate card, the guide
hand-offs, the exit guards and the face-flight overlay, and re-exports
every name it exported. MembersPage.owners.test.tsx pins those exports,
the one-way imports and the owner call order, and
MembersPage.unreadColumn.test.tsx covers the header's unread pills so
the header file clears the per-file coverage floor on its own. Five
source pins follow the code they read: positive halves read the owner
that renders the construct, negative halves read every page file
through one helper, src/test/membersPageFamily.ts.
Moved comments that pointed "above" or "below" at code now in another
file name that file, and the feature map and the crew-mode spec name
the owners.

Part of #20192
@bolichen97
bolichen97 force-pushed the refactor/linecap-f07-memberspage branch from 45027a0 to 5c4a952 Compare October 15, 2026 08:35
@bolichen97

bolichen97 commented Oct 15, 2026 •

Copy link
Copy Markdown
Contributor Author

Rebased: main moved in website/src/test/mobilePanels.compositor.test.ts and docs/feature-map/README.md since this head went green (#20388, ChatPage hook logic into page owners; 83e97cd -> 213339f). The compositor pin keeps reading the whole Crewmates page through readMembersPage() on top of #20388's ChatPage owner join, and the moved comment in useMembersPageDrawers.ts now names drawerTravel's new home, pages/chat/page/sessionsDrawer.ts. No hook added or moved. CI re-running.

@github-actions github-actions Bot added readiness: checking Automated validation is still running and removed merge conflict Branch has merge conflicts with its base — author must resolve before merge readiness: action required A blocking check or review needs attention labels Oct 15, 2026
@bolichen97

Copy link
Copy Markdown
Contributor Author

/ai-review override fable 5c4a952: The Opus lane failed before discovery on the GitHub App API rate limit (run 38706823411); this is infrastructure. Discovery has failed closed on every earlier head of this PR. The Master verified the #20476 and #20388 ports by range-diff.

@github-actions

Copy link
Copy Markdown
Contributor

Human judgment recorded

@bolichen97 marked the fable AI finding as false positive, not applicable, or explicitly accepted for 5c4a952ac59ac5279017f66b320f9a246265f32c.

The Opus lane failed before discovery on the GitHub App API rate limit (run 38706823411); this is infrastructure. Discovery has failed closed on every earlier head of this PR. The Master verified the #20476 and #20388 ports by range-diff.

This decision applies only to this commit. A new push requires a new judgment.

@github-actions github-actions Bot added readiness: action required A blocking check or review needs attention and removed readiness: checking Automated validation is still running labels Oct 15, 2026
@bolichen97

Copy link
Copy Markdown
Contributor Author
  • Base hook order is proven by the recipe, not pinned by a test (answered)

Yes, by the ruling on this PR's earlier finding: a permanent test snapshot of the full hook sequence was judged a tax on every later Crewmates PR and removed, so the order claim is a one-commit proof (the recipe refuses to generate otherwise and reports 209 hook-registering statements identical to base), while the owners test pins the facade's owner call order that later PRs keep.
The proof is reproducible: re-running the recipe on this head's base regenerates the commit byte for byte, and the PR body records the output.

@bolichen97

Copy link
Copy Markdown
Contributor Author
  • Hand check of the Crewmates page against main (answered)

The gateway-free capture harnesses run on main and on this branch gave pixel-identical frames wherever they run (card-rail-exit 10 of 10, panel-ia's first 3), and their later steps fail the same way on main (stale fixtures); the MembersPage specs and the related suite pass unchanged, and every moved body is verbatim.
A maintainer's hand check before merge remains welcome; nothing in this split changes rendered output.

@github-actions github-actions Bot added readiness: passed Eligible automated validation passed for the current revision and removed readiness: action required A blocking check or review needs attention labels Oct 15, 2026

@KiroCrewCaptain KiroCrewCaptain left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Approved by a maintainer after reading the review lanes: every lane is clean on this head, and the automated triage held it only on a size, path, category or doc rule.

@chenmingwei23
chenmingwei23 merged commit 2d77235 into main Oct 15, 2026
103 of 115 checks passed
@chenmingwei23
chenmingwei23 deleted the refactor/linecap-f07-memberspage branch October 15, 2026 15:43
@github-actions github-actions Bot removed the readiness: passed Eligible automated validation passed for the current revision label Oct 15, 2026
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.

4 participants