fix(signals): latest() opens no transaction for its verdict; its reader watches the seam (#4365) by ryansolid · Pull Request #4393 · solidjs/solid

ryansolid · 2026-10-28T21:10:25Z

Summary

Before: a memo or render effect that read latest(a) beside b() computed a torn pass on a batched write.

The latest() read of a's staging opened the flush's transaction for its verdict (provisionalVerdict, joinFuture(null)).
The plain b() read after it then took that manufactured transaction for a hold and was served b's committed value.
Nothing held the flush, so the torn pass was one of its frames. Through a memo, a render effect ran [2,1] before [2,2]. If the torn pass threw, the error stuck.

After: D-22 (clarified 2026-10-28): inside a flush, latest() of a staging that isn't held yet is the frame's read. It serves the staging and opens no hold, and a plain read beside it in the same pass reads the staging too (F-10). If the flush parks, the pass re-derives once as a verdict reader (D-2, A10).

How

provisionalVerdict (verdict.ts): the !probing arm that called joinFuture(null) is gone. A latest() reader goes through watchVerdict like a probe and is marked unanswered, so it re-derives once if the seam parks. Once the flush has joined a transaction, readers route into its verdict lane as before. fix(signals): latest() read inline and through a memo show the same screen #4401's latestOpened is dropped with the arm.
latest() of a flight (verdict.ts pendingVerdict): it joins the verdict lane at once (joinFuture(null) when not probing), as a plain read of a flight does. Both seats serve the committed value (D-22), so the seam has nothing to decide. Without this, true:1 and l1 were asked twice (fix(signals): an async memo over a verdict memo asks each question once #4397's own-window pins).
A verdict effect's staged read before the frame's transaction (core.ts stagedRead): the pass is added to stagedReaders, so the seam re-derives it once if the flush parks. This keeps fix(signals): latest() read inline and through a memo show the same screen #4401's parity screens (v:0 a:0 l:1, inline and memo). It applies to effects only: including verdict memos made async memos ask twice.
Parked flights (lanes.ts laneSeams → adoptParked): when the frame's flight asked the landing's question, the landing adopts it instead of asking again (F-9). When the R5 reader is a sync memo and t is blocked, the frame flights of its async readers are parked too. This fixes m1:220 asked twice when latest() is extracted into a memo (D-25, [2.0 rc.15 + next] Held navigation never publishes delayed pending feedback #4050).
laneSeams lane handling: a verdict lane that a re-derived latest() reader joins doesn't reveal at that seam. A parent whose lane reveals with readers to re-derive lands one round later (revealing), so the lane's answer shows first (F-6). heldBy counts the lanes under t.

Nothing new runs per pass in recompute without the lane engine.

Pins

packages/signals/tests/latest-beside-staged-read-4365.test.ts: the torn pass and the sticking throw (memo and effect readers), both read orders, parked flushes, and display-ahead for a separate async reader. The torn-pass cases fail on next.
packages/signals/tests/latest-two-readers-one-ask-4393.test.ts: each question asked once, and an isPending sibling not changing what v${m0()} shows (D-23, F-6).
verdict-memo-flight-no-reask.test.ts (fix(signals): an async memo over a verdict memo asks each question once #4397): order-only re-pins for R5's frame pass (m0:5 first), each with a comment. Every fix(signals): an async memo over a verdict memo asks each question once #4397 "asked once" pin still holds.
One it.fails: "m1's lane pass is not re-derived over a staging that lands with it (beside)", citing F-6 (amended 2026-10-28) and D-26. With fix(signals): an async memo over a verdict memo asks each question once #4397 merged, m1 is a verdict reader between its flights (D-25), and m0's lane pass re-derives it into the s lane early. The t transaction then lands while m1:1122 is in flight. Telling that pass apart as the hold's work needs a per-pass check, so it is a row for the reader-seat carve. It fails on next too.
Keyed <Show when={latest(count)}> mounts its child twice at a park ([1,3,3]). The second mount is the lane's and is the one shown (boundary-not-born-held-3540, web loading-on-keyed-boundary-3540).

How did you test this change?

At 9c43740, merged with next 9baaba3 (#4383):

signals: 5,978 passed, 3 expected fail;
web (fresh dist): 1,556 passed, including fix(signals): latest() read inline and through a memo show the same screen #4401's latest-memo-inline-parity and every async-gen spec;
tsc --noEmit clean on src;
rules-index current.

Size

The PR doesn't re-record the ledger; it is re-recorded once on next after the batch lands. No floor cap moves. The R5 parked-flight clear in recompute runs only with the lane engine, and ext() no longer seeds the field, so the floors are back to about +0 (-39 B minified against the previous head on every floor, +11 B on the lane builds).

🤖 Generated with Claude Code

https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

…ad of undefined (#4362, #4363, #4376) A memo whose compute reads latest(s) is re-asked as the write's verdict lane work, so it becomes a lane derivation. While its first pass is in flight, nothing is in the lane, and laneValueOf/display fell back to the committed value, which an uninitialized node does not have: - latest(m)/isPending(m) served undefined and probed it pending (#4363); - the lane's end committed that undefined as the node's answer and cleared STATUS_UNINITIALIZED with the flight still up, which tripped the flush loop guard for [latest(m), isPending(m)] (#4362); - in hydration, m2 = fetchC(1 + latest(m1)) asked fetchC(NaN) instead of adopting the streamed answer (#4376). laneValueOf/display now return NOT_PENDING for such a node, so every read throws NotReady (A7; A19 exc. 1) and the lane's end leaves the node in flight. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

Re-measured the store scenario's ledger over next @ 6b792a9. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

…orld (#4365) A memo or render effect that reads `latest(a)` beside `b()` computed a torn pass on a batched write. The `latest()` read of a's staging opened the flush's transaction for its verdict (verdict.ts `provisionalVerdict`). The plain `b()` read after it then took that transaction for a hold (`stagedScreen`): it served b's committed value and made the reader a stale reader of the transaction. Nothing held the flush, so that torn pass was one of the flush's frames. Through a memo, a render effect ran `[2,1]` before `[2,2]`. A throw from the pass stuck, because the landing skips errored readers. A pass seated in the verdict lane only by its own `latest()` read (REACTIVE_PROVISIONAL) now reads a staging as the frame does: it leaves the lane and is recorded in `stagedReaders`. If the frame commits, the pass was the frame's one pass. If the flush parks, the lane seam re-derives it on the committed world as display-ahead work. A reader whose own stale pass threw now re-derives at the landing. `stagedScreen` moves into lanes.ts behind a hook, so the core floor does not carry it. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

…very (async-gen #119, #1238) - verdictValue: read the lane value before routing the reader into the verdict lane, so a node born in the lane with nothing in it yet is read as any uninitialized node (NotReady) without moving its reader (#1238). - reruns: a stale reader whose error is its own (thrown by its pass over the committed value it was served) re-derives at the landing; only an error from an errored source is skipped (#3945). With #4363 no longer committing `undefined`, async-gen case 119 reached this stuck error. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

…-x1ore2-4362

…ments - scripts/size: ledger lines for `signals: + isPending/latest`, the compiled live page and live + router (+46/+46/+47 B minified over next); this PR's store-app line re-measured (97,019 B). - async-gen baseline: 15 recorded failures no longer reproduce (10 undefined-value, 4 final-screen, 1 unhandled-rejection). - RULES-INDEX regenerated; changeset notes the error recovery. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

…, not in reruns The verdict reader's own error from a served committed value (async-gen #119) now re-derives from `dissolveLane` (lane engine only) instead of a widened `reruns` condition, so the floors pay nothing (0 B; the lane scenarios +20 B over the previous form). `reruns` is next's again. Ledger re-measured; async-gen baseline unchanged. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

…er watches the seam (#4365) Rework #4365 to the ruled shape (R5): in provisionalVerdict, latest() no longer calls joinFuture(null). Like a probe read before anything has joined, it goes through watchVerdict. Its pass is the frame's. If the flush parks, it re-derives once in the verdict lane: it is marked unanswered, so the seam does not take the transaction it joined as its answer. A flush that has already joined a transaction (Vl1), and reads outside a flush (Vl3), are unchanged. This drops the stagedScreen variant (REACTIVE_PROVISIONAL, the stagedScreen hook move and the laneSeams change). It also drops this branch's `reruns` change: #4383 handles a self-errored stale reader in lanes.ts dissolveLane, and that branch merges next. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

…stale reader fix Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

… at a park (R-5, #4365) Under design ruling R-5, latest() opens no transaction for its verdict. The keyed condition's first pass is the frame's, and Show mounts A there. The parked flush re-runs the latest() reader once in the verdict lane, and Show mounts A again. The lane's mount is the one shown (the early fallback, still asserted); the held frame's mount is discarded at the lane's reveal. mounts goes [1,2,2] -> [1,3,3] for the static and callback forms over latest(count). The committed-key and `on` rows are unchanged. This re-pin waits on Ryan's ruling. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

changeset-bot · 2026-10-28T21:10:33Z

🦋 Changeset detected

Latest commit: 9c43740

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 15 packages

Name	Type
@solidjs/signals	Patch
test-integration	Patch
@solidjs/web	Patch
@solidjs/babel-plugin	Patch
@solidjs/compiler	Patch
@solidjs/diagnostics	Patch
@solidjs/element	Patch
@solidjs/h	Patch
@solidjs/html	Patch
solid-js	Patch
@solidjs/universal	Patch
board-example	Patch
canvas-example	Patch
server-dbmon-example	Patch
todos-server-example	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

…rror too The re-enqueue of a self-errored stale reader at the lane's end sat inside the non-effect branch, so a render effect whose display-ahead pass threw over a held write's screen value kept its error after the landing (found by the #4365 thread). It now applies to both. Byte-neutral (0 B minified in every scenario); floors 0 B. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

github-actions · 2026-10-28T21:13:34Z

Size (brotli, eager entry chunk)

scenario	head	vs base	minified vs base	minified vs recorded	cap		lazy chunks (not counted)
signals: core floor (createSignal/Memo/Effect/Root/flush)	7.49 KB	0 B	0 B	−11 B	7.44 KB	⚠️ over by 46 B, 31 B minified headroom
signals: + createStore	15.00 KB	−9 B (−0.1%)	0 B	−1235 B	15.04 KB	✅
signals: + isPending/latest	10.99 KB	+280 B (+2.6%)	+830 B	+1218 B	10.38 KB	❌ over by 613 B
signals: + until/refresh (truth readers, no optimism)	8.12 KB	+2 B (+0.0%)	0 B	−9 B	8.05 KB	⚠️ over by 65.99999999999909 B, 29 B minified headroom
app: render + one signal (the simple-app floor)	9.84 KB	+11 B (+0.1%)	0 B	−72 B	9.75 KB	⚠️ over by 94 B, 92 B minified headroom
app: hydrating (no stores) with Show/For/Loading/Errored/lazy	18.72 KB	−12 B (−0.1%)	0 B	+41 B	18.49 KB	❌ over by 234 B	lazy-page.js 0.04 KB
app: hydrating + every store primitive family	32.40 KB	+326 B (+1.0%)	+873 B	+1164 B	31.80 KB	❌ over by 599 B	lazy-page.js 0.04 KB
app: CSR with Show/For/Loading/Errored/lazy	13.49 KB	+6 B (+0.0%)	0 B	+4 B	13.40 KB	⚠️ over by 95 B, 16 B minified headroom	lazy-page.js 0.04 KB
app: CSR, observe tier (same app on the `observe` artifacts)	15.24 KB	+21 B (+0.1%)	0 B	−4 B	15.14 KB	⚠️ over by 102 B, 24 B minified headroom	lazy-page.js 0.04 KB
app: CSR, observe tier + attribution engine enabled	29.83 KB	−5 B (−0.0%)	0 B	−3 B	29.70 KB	⚠️ over by 133 B, 23 B minified headroom	lazy-page.js 0.04 KB
app: compiled floor (one template, one text hole, one delegated click)	10.03 KB	+4 B (+0.0%)	0 B	−71 B	9.93 KB	⚠️ over by 99 B, 91 B minified headroom
app: compiled CSR (JSX todo app: spread/merge/omit, events, class/style, keyed For, Show, Loading + lazy, store)	26.39 KB	+13 B (+0.0%)	0 B	−321 B	26.27 KB	⚠️ over by 125 B, 341 B minified headroom	stats.js 0.18 KB
app: compiled hydrating (the same JSX todo app through hydrate(), compiled hydratable)	32.54 KB	+72 B (+0.2%)	0 B	−220 B	32.25 KB	⚠️ over by 286 B, 240 B minified headroom	stats.js 0.20 KB
frames: eager client consumer (frames client + transport, lazy codec)	11.75 KB	0 B	0 B	0 B	11.76 KB	✅
page: base server components (hydrating + dynamic + frames + sf reference)	33.74 KB	+104 B (+0.3%)	0 B	+120 B	33.37 KB	❌ over by 374 B	assets.js 0.78 KB, decode.js 7.09 KB, lazy-page.js 0.04 KB, regions.js 1.06 KB, trace.js 9.03 KB, wire.js 3.47 KB
page: live server components (base + live/GET + action + isPending/latest)	38.66 KB	+246 B (+0.6%)	+830 B	+1220 B	37.75 KB	❌ over by 915 B	assets.js 0.78 KB, decode.js 7.09 KB, lazy-page.js 0.04 KB, regions.js 1.06 KB, trace.js 9.04 KB, wire.js 3.47 KB
page: compiled base server components (the base page as JSX: templates with class/style/attributes/events, For/Show; no spread)	35.69 KB	+46 B (+0.1%)	0 B	+131 B	35.43 KB	❌ over by 257 B	assets.js 0.78 KB, decode.js 7.09 KB, regions.js 1.06 KB, sc-comments.js 0.20 KB, trace.js 8.99 KB, wire.js 3.47 KB
page: compiled live server components (the compiled base page + live/GET + action + isPending/latest)	42.57 KB	+325 B (+0.8%)	+857 B	+1321 B	41.56 KB	❌ over by 1014 B	eager (counted): web.js 22.66 KB; assets.js 0.78 KB, decode.js 7.09 KB, regions.js 1.07 KB, sc-comments.js 0.19 KB, trace.js 9.05 KB, wire.js 3.47 KB
page: base + router (base page + @solidjs/router: createRouter, two routes, preload, useNavigate)	41.17 KB	+11 B (+0.0%)	0 B	+120 B	40.81 KB	❌ over by 363 B	assets.js 0.78 KB, decode.js 7.09 KB, lazy-page.js 0.04 KB, regions.js 1.06 KB, server.js 1.03 KB, serverForms.js 3.57 KB, trace.js 9.01 KB, wire.js 3.48 KB
page: live + router (live page + @solidjs/router: createRouter, two routes, preload, useNavigate)	48.25 KB	+351 B (+0.7%)	+862 B	+1344 B	47.16 KB	❌ over by 1088 B	eager (counted): client.js 27.83 KB; assets.js 0.78 KB, decode.js 7.09 KB, lazy-page.js 0.04 KB, regions.js 1.07 KB, server.js 1.03 KB, serverForms.js 3.30 KB, trace.js 9.00 KB, wire.js 3.49 KB
server: floor (getRequestEvent + isServer)	1.27 KB	0 B	0 B	−69 B	1.34 KB	✅
server: renderToString (the server-render floor)	22.79 KB	0 B	0 B	0 B	22.80 KB	✅

⚠️ Over the brotli cap within the minified allowance (passes)

signals: core floor (createSignal/Memo/Effect/Root/flush): over brotli cap by 46 B; minified 20,126 B vs 20,137 B recorded with the cap (−11 B) — 31 B of the 20 B minified allowance left; +0 B minified over this PR's base
signals: + until/refresh (truth readers, no optimism): over brotli cap by 65.99999999999909 B; minified 21,937 B vs 21,946 B recorded with the cap (−9 B) — 29 B of the 20 B minified allowance left; +0 B minified over this PR's base
app: render + one signal (the simple-app floor): over brotli cap by 94 B; minified 26,957 B vs 27,029 B recorded with the cap (−72 B) — 92 B of the 20 B minified allowance left; +0 B minified over this PR's base
app: CSR with Show/For/Loading/Errored/lazy: over brotli cap by 95 B; minified 37,211 B vs 37,207 B recorded with the cap (+4 B) — 16 B of the 20 B minified allowance left; +0 B minified over this PR's base
app: CSR, observe tier (same app on the observe artifacts): over brotli cap by 102 B; minified 42,159 B vs 42,163 B recorded with the cap (−4 B) — 24 B of the 20 B minified allowance left; +0 B minified over this PR's base
app: CSR, observe tier + attribution engine enabled: over brotli cap by 133 B; minified 88,498 B vs 88,501 B recorded with the cap (−3 B) — 23 B of the 20 B minified allowance left; +0 B minified over this PR's base
app: compiled floor (one template, one text hole, one delegated click): over brotli cap by 99 B; minified 27,489 B vs 27,560 B recorded with the cap (−71 B) — 91 B of the 20 B minified allowance left; +0 B minified over this PR's base
app: compiled CSR (JSX todo app: spread/merge/omit, events, class/style, keyed For, Show, Loading + lazy, store): over brotli cap by 125 B; minified 80,181 B vs 80,502 B recorded with the cap (−321 B) — 341 B of the 20 B minified allowance left; +0 B minified over this PR's base
app: compiled hydrating (the same JSX todo app through hydrate(), compiled hydratable): over brotli cap by 286 B; minified 100,824 B vs 101,044 B recorded with the cap (−220 B) — 240 B of the 20 B minified allowance left; +0 B minified over this PR's base

❌ Fails the gate

signals: + isPending/latest: over brotli cap by 613 B; minified 30,304 B vs 29,086 B recorded with the cap (+1218 B) — 1198 B past the 20 B minified allowance; +830 B minified over this PR's base. real growth: reduce it, or raise the cap and its recorded minified in this PR with a reason (frozen floors need Size-Exception:)
app: hydrating (no stores) with Show/For/Loading/Errored/lazy: over brotli cap by 234 B; minified 53,661 B vs 53,620 B recorded with the cap (+41 B) — 21 B past the 20 B minified allowance; +0 B minified over this PR's base. real growth: reduce it, or raise the cap and its recorded minified in this PR with a reason (frozen floors need Size-Exception:)
app: hydrating + every store primitive family: over brotli cap by 599 B; minified 98,590 B vs 97,426 B recorded with the cap (+1164 B) — 1144 B past the 20 B minified allowance; +873 B minified over this PR's base. real growth: reduce it, or raise the cap and its recorded minified in this PR with a reason (frozen floors need Size-Exception:)
page: base server components (hydrating + dynamic + frames + sf reference): over brotli cap by 374 B; minified 101,679 B vs 101,559 B recorded with the cap (+120 B) — 100 B past the 20 B minified allowance; +0 B minified over this PR's base. real growth: reduce it, or raise the cap and its recorded minified in this PR with a reason (frozen floors need Size-Exception:)
page: live server components (base + live/GET + action + isPending/latest): over brotli cap by 915 B; minified 116,921 B vs 115,701 B recorded with the cap (+1220 B) — 1200 B past the 20 B minified allowance; +830 B minified over this PR's base. real growth: reduce it, or raise the cap and its recorded minified in this PR with a reason (frozen floors need Size-Exception:)
page: compiled base server components (the base page as JSX: templates with class/style/attributes/events, For/Show; no spread): over brotli cap by 257 B; minified 108,049 B vs 107,918 B recorded with the cap (+131 B) — 111 B past the 20 B minified allowance; +0 B minified over this PR's base. real growth: reduce it, or raise the cap and its recorded minified in this PR with a reason (frozen floors need Size-Exception:)
page: compiled live server components (the compiled base page + live/GET + action + isPending/latest): over brotli cap by 1014 B; minified 125,054 B vs 123,733 B recorded with the cap (+1321 B) — 1301 B past the 20 B minified allowance; +857 B minified over this PR's base. real growth: reduce it, or raise the cap and its recorded minified in this PR with a reason (frozen floors need Size-Exception:)
page: base + router (base page + @solidjs/router: createRouter, two routes, preload, useNavigate): over brotli cap by 363 B; minified 125,507 B vs 125,387 B recorded with the cap (+120 B) — 100 B past the 20 B minified allowance; +0 B minified over this PR's base. real growth: reduce it, or raise the cap and its recorded minified in this PR with a reason (frozen floors need Size-Exception:)
page: live + router (live page + @solidjs/router: createRouter, two routes, preload, useNavigate): over brotli cap by 1088 B; minified 142,327 B vs 140,983 B recorded with the cap (+1344 B) — 1324 B past the 20 B minified allowance; +862 B minified over this PR's base. real growth: reduce it, or raise the cap and its recorded minified in this PR with a reason (frozen floors need Size-Exception:)

_{Bundled with Rolldown (what Vite ships), brotli q11, decimal KB. A scenario fails only when it is over its brotli cap and its minified size is more than 20 B over the minified recorded with the cap; over the cap within that allowance is brotli layout noise and passes with a warning. Caps and their recorded minified in scripts/size/scenarios.js; the floor and page caps in floor-caps.json are frozen (lower only, or Size-Exception: in the PR body). npm run ratchet lowers caps per RC; it never raises one (scripts/size/README.md).}

… re-enqueue

codspeed · 2026-10-28T21:35:46Z

Merging this PR will degrade performance by 7.84%

2 warnings

⚠️ 3 benchmarks are dominated by system calls

System calls cannot be simulated consistently and are left out of the measure. The reported value understates the real cost. Switch to the Walltime instrument to measure them.

Learn more about system calls

See affected benchmarks

⚠️ Base and head ran on different runner types

Some benchmarks with significant performance changes were compared across different runner types, so the changes might not come from the code. Run base and head on the same runner type, or use CodSpeed Macro Runners for more stable results.

Learn more about Macro Runners

Open the report in CodSpeed to investigate

⚡ 1 improved benchmark
❌ 5 regressed benchmarks
✅ 194 untouched benchmarks
⏩ 5 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	`BASE`	`HEAD`	Efficiency
❌	`build`	177.6 µs	228.4 µs	-22.25%
❌	`build + consume`	345.4 µs	393.2 µs	-12.15%
❌	`projection derive: write one NESTED field (reference)`	2.4 ms	2.7 ms	-11.62%
❌	`build + consume`	698.3 µs	745.4 µs	-6.32%
❌	`arrow in template`	11.4 ms	12.1 ms	-5.85%
⚡	`memo + sync render effect only (reference)`	33.4 ms	29.1 ms	+15.04%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

_{Comparing claude/project-thread-x1ore2-4365 (9c43740) with next (9baaba3)}

5 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

ryansolid · 2026-10-28T22:22:50Z

Ruling on the boundary-not-born-held-3540 re-pin: accept [1,3,3].

latest() opens no transaction for its verdict (that is the point of this fix, #4365). When the flush parks, the pass that read latest(count) re-derives once as a verdict reader (D-2, A10); a keyed <Show> whose condition is that read mounts its branch again on that re-derive. That is the cost of keying a mount on a latest() read at a park, not a bug. Do not restore the stagedScreen variant — it puts latest() back to opening the transaction (+110 B) and reintroduces exactly the mechanism this PR removes.

Please note the new expectation in the pin's comment (why 3, with the D-2 / A10 reference) and this PR is good to merge once #4383 lands.

— Claude via Cursor

… a park (D-2, A10) Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

b68942c

Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

ryansolid · 2026-10-28T23:14:54Z

Two checks are red on 03a82df:

job: the same errored-pending-reask-4375.spec.tsx:107 failure fix(signals): latest() of a memo born in a lane throws NotReady, not undefined (#4362, #4363, #4376) #4383 hits, because this PR carries fix(signals): latest() of a memo born in a lane throws NotReady, not undefined (#4362, #4363, #4376) #4383. The cause is a duplicate-re-ask bug already on next, which fix(signals): latest() of a memo born in a lane throws NotReady, not undefined (#4362, #4363, #4376) #4383's NotReady exposes; see the comment on fix(signals): latest() of a memo born in a lane throws NotReady, not undefined (#4362, #4363, #4376) #4383. A fix PR for it is in progress, and I'll bring it in here once it exists.
CodSpeed: projection derive: write one NESTED field (reference) at −11.94%. That benchmark flips ±13–15% with module byte size alone. A V8 lazy sweep after the pre-run gc() either lands in the single measured run or misses it, and a 200 B comment on next reproduced the same jump under callgrind. This PR doesn't touch the store engine, so I'm treating it as noise.

Generated by Claude Code

A memo that reads a verdict memo is a verdict reader (F-6, #4050), and propagateStatus re-derived every verdict reader when its source went pending. Re-derived, an async one was served the committed value it had already derived from (A10) and asked that question again; that flight held the verdict lane, and the source's held landing re-ran it as lane work onto the same committed value, a third ask. Only the landing's re-derive asked the new question. Now an async memo that is a verdict reader only by reading a verdict memo (it opened no isPending/latest window of its own: REACTIVE_WINDOW, set per pass by markVerdictReader) drops CONFIG_VERDICT when that memo goes pending and takes the plain pending mark: it is pending through the flight (F-9) and the hold's work until it lands, reading plainly (D-25, #3884), and asks once at the landing. Sync readers re-derive as before. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

…-x1ore2-4362 Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

…action re-derives once if the flush parks (D-22) In R5, latest() opens no transaction. A render effect that is a verdict reader (`v:${s()} a:${A()} l:${latest(s)}`) read s's staging before the frame had a transaction, so it stayed the frame's pass. Its held run then showed `l:0` until the landing. stagedRead now also lists such an effect for the seam, which re-derives it once as a verdict reader when the flush parks, as D-22 and D-25 require: it shows `v:0 a:0 l:1`. This is limited to effects. A memo is not a leaf, and with async memos the re-run asked its question again. With every memo, the async-gen case e1-3289-1831 ended on [ROOT-E]. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

…at carve (F-6, D-26) With #4397 merged, m1 is a verdict reader between its flights (D-25) and m0's lane pass re-derives it into the s lane early, so the t transaction lands while m1:1122 is in flight. Telling that pass apart as the hold's work needs a per-pass check; it goes to the carve. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

The grouped-vs-split ask order raced its 10 ms flights against a 12 ms sleep on a slow runner (m0b:20 before m1:5). The t write now comes after the s write's first flights land, while m1's question is up. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

The recompute clear runs only with the lane engine, and ext() no longer seeds the field, so builds without lanes drop both (-39 B minified on every floor). Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

…e row (#4411) * test(signals): pin #4402 and #4398 as fixed by #4393; #4399 as a carve row Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2 * test(signals): drive the hold pins on fake timers The change-during-hold pin raced the flights under coverage on CI ([7, 10] for [5, 8, 10]); the script's order now decides it. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2 --------- Co-authored-by: Claude <[email protected]>

The differential proves the table subsumes it: with the mechanism gone the signals, solid and web suites (incl. #4397's and #4393's pins) stay green, the 2k × 3 generator screens match next except the two known closed cases, and the shadow reports 0 classes. The stale-reader mark the same rows fire (ST → _reruns) re-derives the reader at the landing, and holdWork keeps the #4397 wait; the pending-source drop had nothing left to change. The oracle drops the mark with it. + isPending/latest −104 B min; live page −106. Co-authored-by: Claude via Cursor <[email protected]> Co-authored-by: Cursor <[email protected]>

The #4400 case writes s, waits 3 ms, and writes again while a 10 ms ask is in flight. On CI's coverage run the real gap outlived the ask, which landed before the second write and asked `m1:23` (seen on #4455 and #4460). The gaps are part of each scenario, so the file now advances fake timers instead of sleeping. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

…er it lands (#4460) * fix(signals): don't retain the transaction a latest() read opened after it lands `pendingVerdict` (#4453) remembered the flush's transaction a `latest` read of an unheld flight opened in a module-level `latestJoined`, compared only against `flushTransaction` and never cleared: after the seam it kept the last landed transaction (its `_nodes`, `_queues`, `_reruns`) alive until the next such read overwrote it. The mark now lives on the transaction itself (`Transaction.latest`), so it ends with it. Same answer for every probe: a transaction stays the flush's from its opening to the seam, a merge keeps the flush's own, and one merged away is no longer anyone's flush transaction, as before. The property is not `_`-prefixed (as `folded`), so the mangler's assignment is untouched; only bundles importing `latest`/`isPending` change, and they shrink (isPending/latest -13 B brotli, -10 B minified). Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2 * test(signals): the #4393/#4400 two-reader specs run on fake timers The #4400 case writes s, waits 3 ms, and writes again while a 10 ms ask is in flight. On CI's coverage run the real gap outlived the ask, which landed before the second write and asked `m1:23` (seen on #4455 and #4460). The gaps are part of each scenario, so the file now advances fake timers instead of sleeping. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2 --------- Co-authored-by: Claude <[email protected]>

…ns no frame posture (measurement, not for merge) Encoding: REACTIVE_PROBED_HELD / CONFIG_PROBED_HELD (1 << 30, same-bit promotion as REACTIVE_PROBED → CONFIG_VERDICT) set by a probe that finds a held, staged, in-flight or lane's node (verdict.ts `probedHeld`, from `verdictRead`/`watchVerdict`/`pendingVerdict`), inherited through the #4050 arm (core.ts `read`, config at once unless `creatorReads`), and checked where CONFIG_VERDICT alone decided frame/screen reading: `recompute`'s head join, `frameRead`, both `stagedScreen` callers, the pending-node `observeFlight` arm, store.ts's staging arms, lanes.ts `enterLane`/`laneRead`, scheduler.ts zombies, boundaries.ts `on`, and the shadow seat table. Findings (signals suite: 50 pins flip across 15 files; Shape 1 of #4544 does not flip): - D-25 "either read order" is carried by the posture a nothing-found probe leaves: `[h(), isPending(h)]` on the pass where `h` first goes pending or held reads `h` plainly before the probe, and only the earlier probe's posture serves the screen. Under the ruling that pass has no posture and throws/joins (verdict-contract, #4050 "memo [h, inline probe]", #3028, #4393, #4402, #4491, #4504, D-30 store rows, …). A sound encoding needs per-dependency memory ("this pass probes x"), not a node bit. - Once `params` joins T, the `plan` write is held by T (L2); `chrome`'s seam re-derive probes a T-held node and is T's verdict reader; `params` inherits the posture (#4050) / enters T's verdict lane (#4399) and is served the committed `session` again. A "head-joined pass is the hold's work" gate flips Shape 1 but adds #4399/#4408 pins (74 failures). Size (local, base d1d6d51): core floor +1 B br / +3 B min; hello world −5 B br / +3 B min; + isPending/latest +45 B br / +200 B min; compiled live page +75 B br / +250 B min; live + router +151 B br / +254 B min. Co-authored-by: Cursor <[email protected]>

… no caller) Co-authored-by: tembo[bot] <208362400+tembo-io[bot]@users.noreply.github.com>

* refactor(signals): remove the dead latest-opened hold arm (#4393 left no caller) Co-authored-by: tembo[bot] <208362400+tembo-io[bot]@users.noreply.github.com> * chore: drop the scratch coverage configs committed by mistake Co-authored-by: tembo[bot] <208362400+tembo-io[bot]@users.noreply.github.com> --------- Co-authored-by: tembo[bot] <208362400+tembo-io[bot]@users.noreply.github.com>

claude added 11 commits October 28, 2026 18:48

Merge origin/next into claude/project-thread-x1ore2-4365

b4698a5

Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

Merge remote-tracking branch 'origin/next' into claude/project-thread…

69c7aa5

…-x1ore2-4362

Merge #4383 (claude/project-thread-x1ore2-4362) for the self-errored …

47da91e

…stale reader fix Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

Merge #4383 (claude/project-thread-x1ore2-4362) for the render effect…

9b3c2a2

… re-enqueue

Merge next into #4383 and re-record its size ledgers over next @ b68942c

e55631e

Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

ryansolid mentioned this pull request Oct 28, 2026

fix(signals): a frame node re-run outside its hold re-derives the readers of the staging it replaces (#4378) #4390

Merged

claude added 2 commits October 28, 2026 22:23

test(signals, web): note why the keyed Show over latest() mounts 3 at…

044de89

… a park (D-2, A10) Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

Merge #4383 (claude/project-thread-x1ore2-4362) at e55631e, with next @

080b959

b68942c

ryansolid marked this pull request as ready for review October 28, 2026 22:26

claude added 2 commits October 28, 2026 22:34

Merge next into #4383 and re-record its size ledgers over next @ cbc0eeb

3bd1a4d

Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

Merge next into #4393 and re-record its size ledgers over next @ cbc0eeb

03a82df

Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

ryansolid mentioned this pull request Oct 28, 2026

fix(signals): a verdict memo read after a staged read stays the frame's (E9) #4394

Merged

ryansolid mentioned this pull request Oct 29, 2026

fix(signals): a held recovery keeps its shown error for frame and verdict readers #4396

Merged

claude added 3 commits October 29, 2026 01:35

Merge remote-tracking branch 'origin/next' into claude/project-thread…

ad3954b

…-x1ore2-4362 Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

Merge #4397 (verdict memo flight asks once) into #4383

0ae47d1

Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

claude added 5 commits October 29, 2026 09:36

Merge origin/next (#4383 squash) into #4393

f9abd7a

Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

ryansolid mentioned this pull request Oct 29, 2026

[2.0 / next] latest(m2) reveals a lane's answer while m1 of the same lane is still in flight (F-6) #4408

Closed

ryansolid merged commit 81a12d8 into next Oct 29, 2026
7 of 8 checks passed

ryansolid mentioned this pull request Oct 29, 2026

refactor(signals): reader seat — S1 swap (deciders → seat + slot table; lanes.ts split) #4430

Closed

This was referenced Oct 30, 2026

fix(signals): a memo waiting on a held flight is the hold's work at its landing (#4408, #4399) #4510

Merged

test(signals): rule #4417's third mount as R-5's price; S-46 content defect gets its own issue (#4574) #4575

Closed

ryansolid deleted the claude/project-thread-x1ore2-4365 branch November 5, 2026 17:34

ryansolid added a commit that referenced this pull request Nov 10, 2026

refactor(signals): remove the dead latest-opened hold arm (#4393 left…

f074e1a

… no caller) Co-authored-by: tembo[bot] <208362400+tembo-io[bot]@users.noreply.github.com>

ryansolid mentioned this pull request Nov 10, 2026

refactor(signals): remove the dead latest-opened hold arm #4754

Merged

github-actions Bot mentioned this pull request Nov 10, 2026

Release RC packages (rc) #4683

Open

Navigation Menu

Repository navigation

fix(signals): latest() opens no transaction for its verdict; its reader watches the seam (#4365) - #4393

fix(signals): latest() opens no transaction for its verdict; its reader watches the seam (#4365)#4393
ryansolid merged 58 commits into
nextfrom
claude/project-thread-x1ore2-4365

ryansolid commented Oct 28, 2026 •

edited

Loading

Uh oh!

changeset-bot Bot commented Oct 28, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Oct 28, 2026 •

edited

Loading

Uh oh!

codspeed Bot commented Oct 28, 2026 •

edited

Loading

Uh oh!

ryansolid commented Oct 28, 2026

Uh oh!

ryansolid commented Oct 28, 2026

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

ryansolid commented Oct 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

How

Pins

How did you test this change?

Size

Uh oh!

changeset-bot Bot commented Oct 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🦋 Changeset detected

Uh oh!

github-actions Bot commented Oct 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Size (brotli, eager entry chunk)

⚠️ Over the brotli cap within the minified allowance (passes)

❌ Fails the gate

Uh oh!

codspeed Bot commented Oct 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Merging this PR will degrade performance by 7.84%

Performance Changes

Footnotes

Uh oh!

ryansolid commented Oct 28, 2026

Uh oh!

ryansolid commented Oct 28, 2026

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

ryansolid commented Oct 28, 2026 •

edited

Loading

changeset-bot Bot commented Oct 28, 2026 •

edited

Loading

github-actions Bot commented Oct 28, 2026 •

edited

Loading

codspeed Bot commented Oct 28, 2026 •

edited

Loading