Tests grow 2.5 times faster than source: measure the per-fix test weight and the #25911 pilot, then roll the shared-setup merge out to the heaviest components or retire it

director#1 · os-support-ai · session_01WGufzg12BtjFc5rHp7YD1J

Ruled: 6583487846 · Q1 A · Q2 A · 2026-11-02T17:17Z

Filing gate ③, a task the maintainer set. After the director's project assessment reported test lines growing 2.5 times faster than source lines, the maintainer answered 「同意」 to filing this card. ⛔ Not a claim. Reader who acts: the domain:devx seat. Follows #25911 (closed), whose own plan said 「先量 setup 与断言的占比再决定推广」.

Measured (`origin/main`, 2026-10-26 → 11-02, `git log --numstat` over `packages` and `ui`)

class of file	added	removed	net
test files (`.test.`, `.spec.`, `__tests__`, `test/`, `e2e`)	352,859	43,250	+309,609
source	236,290	114,716	+121,574
generated, JSON, fixtures	8,155	3,789	+4,366

tests: 56.5% of spec files are per-issue pins costing 52.5% of suite time — pilot a shared-setup consolidation on the 95 StudioDesignSurface files and split cli-bin.test.ts (74.6 s) #25911 (2026-10-24): 3,263 of 5,773 spec files (56.5 %) were per-issue pins, 52.5 % of measured suite time; the pilot folded StudioDesignSurface.* pins into shared files (PR test(app-shell): fold the other five Interfaces rail label pins onto the shared railLabels harness #26309) and split cli-bin.test.ts (PR test(cli): split cli-bin into two files that share one build behind a lock directory #26333).
Slowest Test Core shard on five recent successful pull_request CI runs (ci.yml, 2026-11-02): 13 min 45 s, 18 min 28 s, 13 min 53 s, 12 min 22 s, 20 min 58 s. ci(test-shards): Test Core's two slowest shards set every core PR's wall clock at ~28 min (5.6 min dependency build + ~20 min tests) while the other six finish in 7–17 — re-derive the cli slicing and shard balance on post-merge readings #25870's fix brought it down from ~28 min; at a net 300,000 test lines a week the suite will grow back.
Example of the per-fix weight: PR fix(app-shell): interface list toolbar honours the userActions spec defaults — Filter, Row height, Group and Hide Fields are on unless written false #29299 changes 20 source lines and adds a 191-line test file.

Ask

Measure, on the last 7 days' merged fix( PRs: test lines added per source line changed (median and p90), and how many new test files each adds.
Measure the tests: 56.5% of spec files are per-issue pins costing 52.5% of suite time — pilot a shared-setup consolidation on the 95 StudioDesignSurface files and split cli-bin.test.ts (74.6 s) #25911 pilot: suite time of the folded StudioDesignSurface.* files before and after.
Report both on this card. If the pilot saved time without losing a pin, roll the shared-setup merge out to the next four components tests: 56.5% of spec files are per-issue pins costing 52.5% of suite time — pilot a shared-setup consolidation on the 95 StudioDesignSurface files and split cli-bin.test.ts (74.6 s) #25911 named (ListView*, ObjectView*, RecordDetailView*, ObjectGantt*), one PR each; if it did not, close the pattern and say so.
⛔ No new gate and no ratio ceiling in CI: a guidance line in the test discipline of AGENTS.md is the most this card may propose, and it goes to the maintainer as a proposal, not a change.

Done when

The two measurements are on this card with their commands, and either the four rollout PRs have merged or the card records why the pattern stops.

triage#1 · os-zhuang · session_01AavokzJ5DndAwitDXvKy4U

Path: none

Triage: p2 · `domain:devx` · `pm:queue` · 2026-11-02T15:57Z

⛔ Not a claim.

Read fresh. A maintainer-set task (「同意」): measure the per-fix test weight and the tests: 56.5% of spec files are per-issue pins costing 52.5% of suite time — pilot a shared-setup consolidation on the 95 StudioDesignSurface files and split cli-bin.test.ts (74.6 s) #25911 pilot, then roll the shared-setup merge out or retire it. Test-infrastructure tooling, so domain:devx as filed. p2.

Generated by Claude Code

director#1 · os-support-ai · session_01WGufzg12BtjFc5rHp7YD1J

Claim: maintainer-directed dispatch · director seat · 2026-11-02T16:30Z
Session: session_01WGufzg12BtjFc5rHp7YD1J
Account: os-support-ai (the seat's linked user as GET /user answers it; the card's assignee)
Branch: claude/issue-29469-test-weight-census
Worktree: objectstack-issue-29469
Domain: domain:devx
Seat: director#1 (the director seat dispatching under the maintainer's direct order; it holds no domain:devx seat)
File surface: none in phase 1 — measurement only, results reported as a comment on this card; ⛔ no source, test or workflow edit in this dispatch (stop on breach; explain in the report)
Container & model: S (measurement, no build of the product), mode:subagent, model: default tier (dispatch-gates.mjs --tier: 「The tier stays the PM's per-card judgment call (floor sonnet · default opus · ceiling fable)」; no path-derived mandate, no file surface), effort: high
Clause-②: no
Metadata-protocol: none
Responsibility: n/a — not a defect card
Thread-read: 6581713880
Premise: holds (the card's figures were measured by this seat at 2026-11-02T15:20Z on origin/main; the dev re-measures them as part of the work)
Serial constraints cleared: none — the dispatch edits no file

Provenance (three pieces): the maintainer, in the director session (session_01WGufzg12BtjFc5rHp7YD1J, claude.ai), verbatim: 「普查测试的卡你现在就可以派发」, after 「同意」 to filing it. The director seat reads 「普查测试的卡」 as this card. The pm-dispatch rule that the director seat never claims backlog yields to the maintainer's order (SKILL.md 〈优先级〉: 维护者裁决 > … > 红线). Triage's first touch 6581713880 (p2, domain:devx, pm:queue) is respected; pm:queue → pm:dispatched in one label write, read back.

Scope of this dispatch: the card's Ask items 1 and 2 (the two measurements) and item 3's verdict, reported on this card. The rollout PRs that item 3 may call for are dispatched afterwards, one per component, by the lane that owns the files (ui/** → domain:ui), each its own claim.

director#1 · os-support-ai · session_01WGufzg12BtjFc5rHp7YD1J

Results: per-fix test weight, the #25911 pilot, and the verdict (Ask 1, 2, 3)

Read at 2026-11-02T16:37Z on objectstack-ai/objectstack at 94d3e2543 (the origin/main tip, committed 2026-11-02T16:10:12Z). Window: first-parent commits on that tip since 2026-10-26T15:20:00Z. That is 1,656 commits, all squash (0 merges). The history is proven present by node scripts/pm/git-history.mjs ensure --since=2026-10-20 (floor 2026-10-17, no fetch needed). Measurement only: no file, PR or label was written.

0. The card's table, re-measured

Same path regexes as the card, with the test class taking precedence over generated, over packages/ and ui/:

class	added	removed	net
test	359,185	43,250	+315,935
source (card's definition)	256,853	142,879	+113,974
generated, JSON, fixtures	8,131	3,789	+4,342

The removed column matches the card exactly. The added column is 2 % (test) and 9 % (source) higher. I did not chase this; the tip moved after the card was read.

The classification moves the multiplier but not the conclusion. The card's "source" includes 22,029 added and 54,985 removed lines that are not code: CHANGELOG.md +16,484 (written by the release tooling), .mdx −23,284, .md −21,705, LICENSE +3,051, and .sh. With source limited to code files (.ts .tsx .js .jsx .mjs .mts .cjs .cts) it is +234,824 / −87,894 / net +146,930. Test against source is then 2.15× on net lines (the card says 2.5×; the raw table gives 2.77×) and 1.53× on added lines. Tests still outgrow source by about 2 to 1.

1. Per-fix test weight (Ask 1)

Population: the 816 first-parent commits in the window whose subject starts fix( (2 more start fix: or fix! and are not counted). Each file is classified as: test (card regex), generated (card regex), prose (.md, .mdx, .txt, LICENSE*, which includes .changeset/), or source (everything else, any path). Source lines changed = added + removed.

reading	median	p90	other
test lines added ÷ source lines changed (809 fixes with source over 0)	2.28	7.47	p75 4.07, mean 3.59; aggregate 189,425 ÷ 99,540 = 1.90
test lines added per fix	196	476	total 189,425
source lines changed per fix	78	266	total 99,540
new test files per fix	1	2	876 files over 816 fixes, mean 1.07

7 fixes change no source line, and 2 of those add tests. Counting them as infinite moves the median to 2.30.
83.4 % of fixes add more test lines than the source lines they change.
69.7 % of fixes add at least one new test file. 78.8 % of fix test lines (149,189) land in new files rather than in existing ones.
fix( commits wrote 52.5 % of every test line added in the window (189,425 of 360,886, repo-wide).
p90 is nearest-rank. The card's example (fix(app-shell): interface list toolbar honours the userActions spec defaults — Filter, Row height, Group and Hide Fields are on unless written false #29299: 20 source lines, a 191-line test file) is a typical fix, not an outlier.

2. The pilot (Ask 2)

Instruments and their radius

(a) CI before: ui/scripts/test-shard-timings.json. This is the per-file vitest module diagnostic (environment + prepare + setup + import + run) from CI, median of 6 full-battery runs on 2026-10-19T06:37:09Z, keyed by pathKey() (sha1 of the ui-relative path, first 16 hex). To check the keying: cli-bin.test.ts reads 74.12 s, which is the 74.1 s test(cli): split cli-bin into two files that share one build behind a lock directory #26333 cites.
(b) Local before and after: the authors' own A/B in the bodies of PR test(app-shell): fold ten Interfaces rail label pins onto one shared StudioDesignSurface harness #26255 and PR test(app-shell): fold the other five Interfaces rail label pins onto the shared railLabels harness #26309. Same instrument (module diagnostics), same box, same command (vitest run packages/app-shell/src/views/studio-design/StudioDesignSurface --maxWorkers=2), shared-box seconds. I did not re-run it, because this dispatch is a no-build container.
(c) CI after: NOT MEASURED. The per-file timings reporter was not ported when ui/ was imported (see the comment above Run the ui tests in ci.yml), and the dataset has not been refreshed since 2026-10-19T06:37:09Z. No CI per-file reading exists after the fold.

reading	before	after
the ten folded pins, CI (a)	10 files, 70.5 s	not measured
the ten folded pins, local (b)	10 files, 76.6 s CPU (38.1 + 38.5)	1 file, 13.8 s CPU (−82 %)
`StudioDesignSurface.*` family, local (b)	95 files, 1,022.1 s CPU, 528.4 s wall	86 files, 954.9 s CPU, 493.8 s wall (−6.6 % CPU, −6.5 % wall)

The CI before (70.5 s) and the local before (76.6 s) agree within 9 %, so the local scale holds. Each folded file saved about 6.5 s of fixed cost, and the folded file's own fixed cost stayed about 7.5 s.

No pin lost (my own check, from git only). The ten original files, read at the parent of each fold (697a307a2^ and 3249e204a^), hold 50 it/test titles and 153 expect( calls. StudioDesignSurface.railLabels.test.tsx at 3249e204a holds the same 50 titles as an identical set, and 153 expect( calls. At the tip it holds 54 titles and 172 expect( calls, and 0 of the original titles are missing. The 4 extra titles came from later commits that appended to the shared file (for example #27039 and #28825).

The cli-bin split (#26333) is a split, not a fold. The heaviest file went from 87.2 s to 57.5 s cold wall, the spawn CPU is unchanged, and CI CPU rises by one build when the two files land on different legs. It is not evidence for or against the shared-setup merge.

Inflow after the pilot. StudioDesignSurface.* test files: 95 before, 86 after #26309, 116 at the tip. That is 30 new per-issue pin files in the 7 days since, and 0 deleted. At about 6.5 s of fixed cost each, that is roughly 195 s, about 3× the 67 s the fold saved.

The four components #25911 named, counted by test-file basename COMPONENT.*.test.ts(x):

family	files at `f143cd88` (pilot base)	files at tip	CI module time at 2026-10-19 (a)
`StudioDesignSurface`	95	116	948.0 s
`ListView`	110	128	615.7 s
`ObjectView`	131	141	804.8 s
`RecordDetailView`	70	99	402.0 s
`ObjectGantt`	53	56	274.3 s

3. Verdict (Ask 3): roll out, and land new pins in the shared file

Roll out. The pilot cut its ten pins from 76.6 s to 13.8 s of CPU (the family by 6.6 %) and kept 50 of 50 titles and 153 of 153 expect( calls, which meets the card's bar.

Folding alone does not keep up with the inflow. The same family gained 30 new pin files in the following week, about 3× the time the fold saved, and 78.8 % of fix test lines go into new files. A rollout PR pays off only if later pins for that component go into its shared file.

So, under Ask 4 (a proposal to the maintainer, not a change, and no gate or ceiling): add one guidance line to the test discipline in AGENTS.md: "A new pin for a component that already has a shared harness file goes in as a top-level describe in that file, not as a new file." The four rollout PRs (ListView*, ObjectView*, RecordDetailView*, ObjectGantt*) are for the owning lane (domain:ui) to dispatch.

The rollout has no CI after-reading until the per-file timings reporter and its refresh workflow are ported into ci.yml. Today each rollout PR can only cite a local A/B like #26309's.

Exact commands (scripts are in the dev's scratchpad, read-only over git)

# SINCE = the window start declared at the top of this comment
git fetch origin main
git worktree add --no-track ../objectstack-issue-29469 -b claude/issue-29469-test-weight-census origin/main
node scripts/pm/git-history.mjs ensure --since=2026-10-20
# table + per-fix: one numstat pass, parsed in Python (renames resolved to the new path, binary rows dropped)
git log 94d3e2543 --first-parent --since="$SINCE" --numstat --format='@@%H%x09%cI%x09%s'
# new files per commit
git log 94d3e2543 --first-parent --since="$SINCE" --diff-filter=A --name-only --format='@@%H'
# test regex:  \.test\.|\.spec\.|__tests__|/test/|/tests/|e2e
# gen regex:   generated|\.snap|\.json|baseline|fixture   (test wins on overlap)
# pilot, CI before: sha1(ui-relative path)[:16] looked up in ui/scripts/test-shard-timings.json .files
# pilot, pins kept: git show REV:PATH for the 10 originals at 697a307a2^ / 3249e204a^ and the folded file at 3249e204a / 94d3e2543,
#   counting it/test titles and expect( calls
# inflow: git ls-tree -r --name-only REV ui/ | basename COMPONENT.*.test.ts(x), at f143cd88, 3249e204a, 94d3e2543
git log 3249e204a..94d3e2543 --first-parent --diff-filter=A --name-only --format= -- 'ui/packages/app-shell/src/views/studio-design/StudioDesignSurface.*.test.tsx'

director#1 · os-support-ai · session_01WGufzg12BtjFc5rHp7YD1J
os-dev-report

{
"issue": 29469,
"status": "done",
"branch": "claude/issue-29469-test-weight-census",
"pr": null,
"session": "session_01WGufzg12BtjFc5rHp7YD1J",
"premise_still_valid": true,
"summary": "Measurement only, results comment 6582607393 on #29469. Ask 1: 816 fix( commits; test lines added / source lines changed median 2.28, p90 7.47; new test files per fix median 1, p90 2 (69.7% add one; 78.8% of fix test lines land in new files). Ask 2: the ten folded pins went 76.6 s to 13.8 s CPU (local A/B from PR #26255/#26309 bodies; CI before 70.5 s from ui/scripts/test-shard-timings.json; CI after NOT MEASURED, per-file reporter not ported), family -6.6% CPU, 50/50 titles and 153/153 expect( kept. Verdict: roll out, with a proposed AGENTS.md guidance line that new pins land in the shared file, because StudioDesignSurface.* gained 30 pin files in the week after the fold (about 3x the saving). Card table re-measured: code-only net test:source is 2.15x, not 2.5x (card's source class carries CHANGELOG/md/mdx/LICENSE). Writes: 2 comments via post-stamped relay; mcp_calls 0.",
"tests": "No product test run (no-build dispatch). Measurements: git log TIP --first-parent --since=SINCE --numstat parsed in Python at 94d3e25 (1,656 commits, 0 merges); history proven by git-history.mjs ensure --since=2026-10-20 (exit 0, floor 2026-10-17). Pin-kept check: 10 originals at 697a307^/3249e204a^ = 50 titles, 153 expect(; folded file at 3249e20 = same 50-title set, 153 expect(; at tip 54/172, 0 originals missing. CI-dataset keying control: cli-bin.test.ts = 74.12 s, matching the 74.1 s #26333 cites. No ablation: no tested behaviour changed.",
"gates": "No diff, so no derived gate family and no PR check-runs; dispatch-gates.mjs not run (nothing to derive from). post-stamped dry-run exit 2 once (MIXED stamp refusal on a bare --since stamp in the commands block), fixed to a SINCE variable, dry-run exit 0, post exit 0 with byte-identical read-back (8668 bytes).",
"dogfood_battery": "n/a — no PR, no ! title and no (narrowing) arm",
"mcp_calls": "0",
"api_writes": "2 — both through scripts/pm/post-stamped.mjs on the fleet-write relay (repository_dispatch then POST /repos//issues/29469/comments): the results comment 6582607393 and this os-dev-report comment. No git push (no edit, so no empty-branch probe), no PR, no label.",
"open_questions": [
{
"question": "Accept the AGENTS.md test-discipline guidance line proposed under Ask 4 (a new pin for a component that already has a shared harness file goes in as a top-level describe in that file, not a new file)?",
"options": [
"A propose it to the maintainer as written, as a separate docs PR after the ruling",
"B roll out the four folds without the line"
],
"recommendation": "A, because the StudioDesignSurface family gained 30 pin files in the 7 days after the fold, about 3x the CPU the fold saved, so folds without the landing rule are outrun within a week."
},
{
"question": "Port the ui per-file timings reporter and its refresh workflow into ci.yml so a rollout has a CI after-reading?",
"options": [
"A file it as a devx card before the rollout PRs",
"B let each rollout PR cite a local A/B like #26309"
],
"recommendation": "A, because no CI per-file reading has existed since the ui/ import, so neither the pilot nor the rollout can be checked in CI."
}
],
"security_findings": "none",
"out_of_scope_findings": [
"carrier: PM (director#1) · the ui per-file timings reporter was not ported with the ui/ import (ci.yml comment above Run the ui tests says so), so ui/scripts/test-shard-timings.json is frozen at its 2026-10-19 measurement; observation, noted and not filed · dedupe words: test-file-timings reporter, shard timings refresh, ui per-file timings, test-shard-timings.json stale"
]
}

director#1 · os-support-ai · session_01WGufzg12BtjFc5rHp7YD1J

Collected and accepted · measurement dispatch (claim 6582445213) · 2026-11-02T16:41Z

Verdict: ACCEPT. Results 6582607393 and report 6582630893 are on this card, both through the relay, read back. No PR, no file edit, mcp_calls 0: the dispatch's shape held.
Seat spot checks, own instrument: fix( first-parent subjects on origin/main 2026-10-26T15:20Z → 2026-11-02 15:20 UTC = 811 (the dev's 816 runs to the 16:10Z tip); net test lines ÷ net code-source lines (.ts/.tsx/.js/.mjs/.cjs/.jsx, generated and fixtures excluded) over packages + ui = 313,791 ÷ 145,358 = 2.16 (the dev: 2.15). The body's "2.5 times" counted CHANGELOG and prose lines as source; the corrected figure is about 2.15. The conclusion stands.
Item 3 verdict: roll out the shared-setup fold to ListView*, ObjectView*, RecordDetailView*, ObjectGantt* (pilot: the ten folded pins 76.6 s → 13.8 s CPU, local A/B; no title or expect( lost). The rollout PRs are domain:ui work and are not dispatched until the two questions below are ruled, because the first decides whether a fold stays folded.

Two questions for the maintainer (the dev's `open_questions`, analysed by this seat)

Q1 — a test-discipline line in AGENTS.md: 「a new pin for a component that already has a shared harness file goes in as a top-level describe in that file, not as a new file.」 Reading behind it: StudioDesignSurface.* went 95 → 86 files with the fold, then 116 at the tip — 30 new per-issue pin files in 7 days, about 3× what the fold saved. 78.8 % of fix test lines land in new files.

A add the line (one docs PR, governed text, your approval). B roll out without it.

Q2 — the ui/ per-file timings reporter, not ported when ui/ was imported: ui/scripts/test-shard-timings.json has been frozen since 2026-10-19, so neither the pilot nor a rollout can be read in CI, and Test UI shard balancing runs on week-old data.

A file one domain:devx card to port the reporter and its refresh into ci.yml (Tier H, your approval) before the rollout. B each rollout PR cites a local A/B as test(app-shell): fold the other five Interfaces rail label pins onto the shared railLabels harness #26309 did.

四轴分析

项目长远合理性:Q1 选 A。按组件共用一个测试文件、用 describe 分块,是 Jest/Vitest 生态的常规组织;不加这条规则,合并一次、一周又长回来,实测已是 3 倍。Q2 选 A。CI 自己测出的逐文件耗时,才是分片均衡和测试瘦身的长期依据;本地 A/B 每次换机器、换口径。
实际业务拉动:有。修复 PR 中位数每改 1 行源码配 2.28 行测试,七成修复新开测试文件;Test Core 现在 12–21 分钟,增长会把它推回去。
防 AI 犯错:Q1 是给 AI 读的约定,告诉它新 pin 写进哪里,AI 最常见的做法正是「每张卡一个新文件」。这类约定不属于「声明即强制」的范围,不需要门禁。Q2 不直接防错,但让回归可见。
创业阶段不扩散:Q1 只加一行规则文字,不加门禁。Q2 恢复一个导入时丢掉的测量工具,不是新增门禁;代价是一次 workflow 改动和您一次批准。

os-decision-facets

① 项目长远合理性:Q1 A 按组件共用测试文件是生态常规,不加规则一周即长回;Q2 A 以 CI 实测取代每次换口径的本地 A/B。
② 实际业务拉动:有——修复中位 2.28 行测试/行源码,七成新开文件,StudioDesignSurface 一周新增 30 个 pin 文件。
③ 防 AI 犯错:Q1 直接改掉 AI「一卡一新文件」的默认做法;Q2 让回归在 CI 可见。
④ 创业阶段不扩散:Q1 一行约定无门禁;Q2 恢复导入时丢失的测量,非新增门禁,一次 workflow 改动。
Prior rulings read: "per-issue pin" "shared harness" "test-shard-timings" → tests: 56.5% of spec files are per-issue pins costing 52.5% of suite time — pilot a shared-setup consolidation on the 95 StudioDesignSurface files and split cli-bin.test.ts (74.6 s) #25911 (pilot), ci(test-shards): Test Core's two slowest shards set every core PR's wall clock at ~28 min (5.6 min dependency build + ~20 min tests) while the other six finish in 7–17 — re-derive the cli slicing and shard balance on post-merge readings #25870 (Test Core shards, on hold); AGENTS.md test discipline: no file-placement rule; thread: none

自检:只看①选 Q1 A、Q2 A;②③④ 是否翻转:否。

推荐:Q1 A,Q2 A。 Q1 B 不取:实测一周内长回 3 倍。Q2 B 不取:本地 A/B 不能复查,也不喂分片均衡。置信缺口: 报告器移植的工作量未测(原实现在 objectui 仓,导入时未带过来)。

Analysis-tier: claude-opus-5-5 (served at analysis; not CONTRACT_REVIEW_TIER).

State: the measurement claim is released. pm:dispatched → needs-user-decision in one label write, assignee cleared; this card is the anchor for the ruling. After the ruling: Q1 A → one docs PR in the skills lane quoting it; Q2 A → one domain:devx card; then the four rollout cards in domain:ui.

director#1 · os-support-ai · session_01WGufzg12BtjFc5rHp7YD1J

Brief-claimed-by: session_01WGufzg12BtjFc5rHp7YD1J · batch #458 · 2026-11-02T17:11Z

⛔ Not a claim on the work. The two questions in 6582704041 are batch #458 item 1, presented in the director session.

director#1 · os-support-ai · session_01WGufzg12BtjFc5rHp7YD1J

Ruling: batch #458 item 1 · Q1 letter A · Q2 letter A · maintainer 「同意」 2026-11-02T17:15Z

Record: the director presented the two questions of 6582704041 in the director session with A recommended for both; the maintainer answered 「同意」. Brief-claim 6583378464.
Q1 A: one line in the test discipline of AGENTS.md: a new pin for a component that already has a shared harness file goes in as a top-level describe in that file, not as a new file. Workload: domain:skills · 1 docs PR (governed text, the maintainer's approval) · no precedent.
Q2 A: port the ui/ per-file timings reporter and its dataset refresh into ci.yml, so CI can read the rollout. Workload: domain:devx · 1 PR touching .github/workflows/** (Tier H, the maintainer's approval) · effort unmeasured (the original lives in the objectui repository and was not carried over).
Then: the shared-setup fold rolls out to ListView*, ObjectView*, RecordDetailView*, ObjectGantt*, one PR each, domain:ui.
Exit stroke on this card (needs-user-decision → working state) and the Q2 card are the domain:devx seat's; the Q1 PR is the domain:skills seat's. The director files both cards in this summon and knocks both lanes.

domain:devx#1 · os-warren · session_01JQwxbaBjisxkQo9hht9bFB

Exit stroke for ruling 6583487846 (Q1 A, Q2 A), done by the domain:devx seat the ruling names. needs-user-decision → pm:blocked in this stroke.

Ruling-ref: 6583487846
Blocked-by: #29517
Blocked-by: #29516

#29517 (Q2: port the Test UI per-file timings reporter and dataset refresh into ci.yml, domain:devx, Tier H) and #29516 (Q1: the shared-harness line in AGENTS.md, domain:skills, governed) are the two cards the director filed. This card stays open as the anchor for the rollout that follows them: the shared-setup fold to ListView*, ObjectView*, RecordDetailView*, ObjectGantt*, one PR each in domain:ui, read in CI through #29517's timings.

Navigation Menu

Repository navigation

Tests grow 2.5 times faster than source: measure the per-fix test weight and the #25911 pilot, then roll the shared-setup merge out to the heaviest components or retire it #29469

Measured (`origin/main`, 2026-10-26 → 11-02, `git log --numstat` over `packages` and `ui`)

Ask

Done when

Triage: p2 · `domain:devx` · `pm:queue` · 2026-11-02T15:57Z

Results: per-fix test weight, the #25911 pilot, and the verdict (Ask 1, 2, 3)

0. The card's table, re-measured

1. Per-fix test weight (Ask 1)

2. The pilot (Ask 2)

3. Verdict (Ask 3): roll out, and land new pins in the shared file

Collected and accepted · measurement dispatch (claim 6582445213) · 2026-11-02T16:41Z

Two questions for the maintainer (the dev's `open_questions`, analysed by this seat)

四轴分析

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Tests grow 2.5 times faster than source: measure the per-fix test weight and the #25911 pilot, then roll the shared-setup merge out to the heaviest components or retire it #29469

Description

Measured (origin/main, 2026-10-26 → 11-02, git log --numstat over packages and ui)

Ask

Done when

Activity

objectstack-fleet commented on Nov 2, 2026

Triage: p2 · domain:devx · pm:queue · 2026-11-02T15:57Z

objectstack-fleet commented on Nov 2, 2026

objectstack-fleet commented on Nov 2, 2026

Results: per-fix test weight, the #25911 pilot, and the verdict (Ask 1, 2, 3)

0. The card's table, re-measured

1. Per-fix test weight (Ask 1)

2. The pilot (Ask 2)

3. Verdict (Ask 3): roll out, and land new pins in the shared file

objectstack-fleet commented on Nov 2, 2026

objectstack-fleet commented on Nov 2, 2026

Collected and accepted · measurement dispatch (claim 6582445213) · 2026-11-02T16:41Z

Two questions for the maintainer (the dev's open_questions, analysed by this seat)

四轴分析

objectstack-fleet commented on Nov 2, 2026

objectstack-fleet commented on Nov 2, 2026

objectstack-fleet commented on Nov 2, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

Measured (`origin/main`, 2026-10-26 → 11-02, `git log --numstat` over `packages` and `ui`)

Triage: p2 · `domain:devx` · `pm:queue` · 2026-11-02T15:57Z

Two questions for the maintainer (the dev's `open_questions`, analysed by this seat)