[ryansolid]

Summary

Before: a sync memo that threw, then recovered under an open action, lost its error from the screen early. createErrorBoundary(() => "X" + latest(s) + m()), with m throwing for the committed s and recovering for the value an open action writes, showed X1undefined while the action was still open. The same thing happened when the read went through a memo, or through <Errored>{"X" + latest(s) + m0()}</Errored> in JSX. Expected per the docs: the error stays the display until the frame that carries its replacement reveals (D-39, D-41; F-7 for content derived from a held change). A verdict or frame reader beside latest() reads the screen (D-25, A10), and the screen still shows the error.

After: the boundary shows the error until the action commits, then shows X11. No pass of the content computes …undefined.

The staged recovery cleared the memo's error status without setting _shownError, so a reader served the committed value got the empty _value behind the error.

How: recompute keeps the error it replaces as _shownError when a recovery is staged; the commit already clears it. read serves a frame reader of a held node the way it serves an untracked read, and the verdict arm of serve does the same. Both throw a shown error (throwShown), as an untracked read already did. At the landing, reruns re-derives an errored stale reader whose error's source is no longer errored. The #3945 skip still applies while that source is errored. The setting is limited to the staged branch on purpose: when it was set before the equality gate, a boundary tree that recovered to an equal value (loading) was forced into a change, and #2701 broke. Recovering to a value equal to the last good one under a hold is still silent; that was already true on next and this PR leaves it alone. Offsets, all perf-neutral, pay for most of the floor cost:

• recompute captures its outgoing error and pending sources with && instead of ? : undefined, and notifyStatus/propagateStatus do the same for the pending source. Every reader of these values is truthy-guarded.
• notifyStatus and caught drop a trailing undefined argument.
• _shownError resets to null.

Review fix (liveness): a memo the content created on its first pass and kept in a closure ("X" + latest(s) + (cached ||= createMemo(...))()) is disposed with the pass that created it. Its staged recovery throws the shown error, and that error names the dead memo as its source. The error boundary recorded the dead memo, the seam dropped it at once (#4384 drops an entry whose reader or source is disposed), and the output re-derived, re-caught and re-recorded it until the flush threw "Potential Infinite Loop Detected". caught now waits on the reader when the source that threw is disposed. The reader stays errored until the landing re-derives it (reruns), then the boundary shows X11. reruns also guards an error with no source (?.source?. defaults to errored, so the #3945 skip still applies).

I also tried moving the verdict arm of serve behind a hook that verdict.ts installs. It cut the floors by about 22 B, but the hook's wrapper and install code cost every isPending/latest build about +60 B and broke the live server-components floor, so I dropped it.

How did you test this change?

• packages/signals/tests/error-staged-recovery.test.ts: the read inline and through a memo, with the action's flush, a mainline write that re-runs the view over the held recovery, and the landing. It fails on next (X1undefined). It also pins the review's closure-kept memo (inline and through a memo), which looped on the previous head.
• packages/web/test/errored-held-recovery-latest.spec.tsx: <Errored> with "X" + latest(s) + m0() under an open action (inline and through a memo), plus the review's lazy-cached shape: <Errored>{"X" + latest(s) + (cached ||= createMemo(...))()}</Errored>.
• Every case from the review's scratch files (signals A–K, web component/lazy-cached/isPending/span-leaf/grouped) passes, with no undefined on screen and no loop.
• After a clean npx turbo build --force of signals, solid-js and web, merged with next @ 957f065:
  • signals: 384 files, 5867 passed.
  • solid: 917 passed.
  • web default: 1537 passed. Hydrate: 686 passed. Server: 1892 passed, plus the one known env failure (server-functions-adapter-request).
• CI's earlier failure on e36a6b9 was attribution-waterfall-eval.test.ts "FINDING: one scope reading two async sources serializes them" timing out at 5 s. That test is timing-based (sleep(12) chains) and passes locally in about 40 ms across 15 repeated runs. The same timeout also hit CI on two unrelated branches on 2026-10-27 (feat/low-priority-modulepreload, claude/project-thread-x1ore2-4322). It is a pre-existing flake, not this change.
• Size: on request, this PR records no ledger lines and raises no caps. The ledger is re-recorded on next after the batch merge. Measured locally against next @ 957f065: the core floor is +15 B minified and the compiled floor (hello world) is 9,987 B brotli (+25). gate.mjs flags recorded-minified growth on three scenarios: hydrating + every store family +28, compiled live server components +36, live + router +36. check-floor-caps.mjs origin/next reports no cap raised.

https://claude.ai/code/session_015hPxMkC3jg3Uivh65gvaf2

---

Generated by Claude Code

[changeset-bot]

🦋 Changeset detected

Latest commit: da52352

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

[github-actions]

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)	7.45 KB	+16 B (+0.2%)	+15 B	−135 B	7.44 KB	⚠️ over by 8 B, 155 B minified headroom
signals: + createStore	14.96 KB	+21 B (+0.1%)	+14 B	−1359 B	15.04 KB	✅
signals: + isPending/latest	10.40 KB	+27 B (+0.3%)	+22 B	−18 B	10.38 KB	⚠️ over by 23 B, 38 B minified headroom
signals: + until/refresh (truth readers, no optimism)	8.07 KB	+24 B (+0.3%)	+14 B	−140 B	8.05 KB	⚠️ over by 17.99999999999909 B, 160 B minified headroom
app: render + one signal (the simple-app floor)	9.79 KB	+20 B (+0.2%)	+14 B	−196 B	9.75 KB	⚠️ over by 41 B, 216 B minified headroom
app: hydrating (no stores) with Show/For/Loading/Errored/lazy	18.70 KB	+14 B (+0.1%)	+25 B	−136 B	18.49 KB	⚠️ over by 212 B, 156 B minified headroom	lazy-page.js 0.04 KB
app: hydrating + every store primitive family	31.89 KB	+96 B (+0.3%)	+25 B	+25 B	31.80 KB	❌ over by 94 B	lazy-page.js 0.04 KB
app: CSR with Show/For/Loading/Errored/lazy	13.42 KB	−6 B (−0.0%)	+28 B	−173 B	13.40 KB	⚠️ over by 20 B, 193 B minified headroom	lazy-page.js 0.04 KB
app: CSR, observe tier (same app on the observe artifacts)	15.19 KB	+23 B (+0.2%)	+28 B	−181 B	15.14 KB	⚠️ over by 45 B, 201 B minified headroom	lazy-page.js 0.04 KB
app: CSR, observe tier + attribution engine enabled	29.78 KB	+29 B (+0.1%)	+28 B	−180 B	29.70 KB	⚠️ over by 76 B, 200 B minified headroom	lazy-page.js 0.04 KB
app: compiled floor (one template, one text hole, one delegated click)	9.99 KB	+25 B (+0.3%)	+14 B	−195 B	9.93 KB	⚠️ over by 57 B, 215 B minified headroom
app: compiled CSR (JSX todo app: spread/merge/omit, events, class/style, keyed For, Show, Loading + lazy, store)	26.30 KB	+77 B (+0.3%)	+28 B	−498 B	26.27 KB	⚠️ over by 25 B, 518 B minified headroom	stats.js 0.18 KB
app: compiled hydrating (the same JSX todo app through hydrate(), compiled hydratable)	32.43 KB	−41 B (−0.1%)	+31 B	−397 B	32.25 KB	⚠️ over by 179 B, 417 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)	33.66 KB	+73 B (+0.2%)	+25 B	−57 B	33.37 KB	⚠️ over by 293 B, 77 B minified headroom	assets.js 0.78 KB, decode.js 7.09 KB, lazy-page.js 0.04 KB, regions.js 1.06 KB, trace.js 9.02 KB, wire.js 3.47 KB
page: live server components (base + live/GET + action + isPending/latest)	38.05 KB	+17 B (+0.0%)	+27 B	+13 B	37.75 KB	⚠️ over by 295 B, 7 B minified headroom	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: compiled base server components (the base page as JSX: templates with class/style/attributes/events, For/Show; no spread)	35.60 KB	−18 B (−0.1%)	+19 B	−52 B	35.43 KB	⚠️ over by 169 B, 72 B minified headroom	assets.js 0.78 KB, decode.js 7.09 KB, regions.js 1.06 KB, sc-comments.js 0.20 KB, trace.js 9.00 KB, wire.js 3.47 KB
page: compiled live server components (the compiled base page + live/GET + action + isPending/latest)	41.90 KB	−65 B (−0.2%)	+33 B	+33 B	41.56 KB	❌ over by 341 B	eager (counted): web.js 22.43 KB; assets.js 0.78 KB, decode.js 7.09 KB, regions.js 1.07 KB, sc-comments.js 0.19 KB, trace.js 9.09 KB, wire.js 3.47 KB
page: base + router (base page + @solidjs/router: createRouter, two routes, preload, useNavigate)	41.20 KB	+176 B (+0.4%)	+22 B	−60 B	40.81 KB	⚠️ over by 392 B, 80 B minified headroom	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)	47.56 KB	+18 B (+0.0%)	+33 B	+33 B	47.16 KB	❌ over by 405 B	eager (counted): client.js 27.57 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 9.01 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 8 B; minified 20,002 B vs 20,137 B recorded with the cap (−135 B) — 155 B of the 20 B minified allowance left; +15 B minified over this PR's base
• signals: + isPending/latest: over brotli cap by 23 B; minified 28,675 B vs 28,693 B recorded with the cap (−18 B) — 38 B of the 20 B minified allowance left; +22 B minified over this PR's base
• signals: + until/refresh (truth readers, no optimism): over brotli cap by 17.99999999999909 B; minified 21,806 B vs 21,946 B recorded with the cap (−140 B) — 160 B of the 20 B minified allowance left; +14 B minified over this PR's base
• app: render + one signal (the simple-app floor): over brotli cap by 41 B; minified 26,833 B vs 27,029 B recorded with the cap (−196 B) — 216 B of the 20 B minified allowance left; +14 B minified over this PR's base
• app: hydrating (no stores) with Show/For/Loading/Errored/lazy: over brotli cap by 212 B; minified 53,484 B vs 53,620 B recorded with the cap (−136 B) — 156 B of the 20 B minified allowance left; +25 B minified over this PR's base
• app: CSR with Show/For/Loading/Errored/lazy: over brotli cap by 20 B; minified 37,034 B vs 37,207 B recorded with the cap (−173 B) — 193 B of the 20 B minified allowance left; +28 B minified over this PR's base
• app: CSR, observe tier (same app on the observe artifacts): over brotli cap by 45 B; minified 41,982 B vs 42,163 B recorded with the cap (−181 B) — 201 B of the 20 B minified allowance left; +28 B minified over this PR's base
• app: CSR, observe tier + attribution engine enabled: over brotli cap by 76 B; minified 88,321 B vs 88,501 B recorded with the cap (−180 B) — 200 B of the 20 B minified allowance left; +28 B minified over this PR's base
• app: compiled floor (one template, one text hole, one delegated click): over brotli cap by 57 B; minified 27,365 B vs 27,560 B recorded with the cap (−195 B) — 215 B of the 20 B minified allowance left; +14 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 25 B; minified 80,004 B vs 80,502 B recorded with the cap (−498 B) — 518 B of the 20 B minified allowance left; +28 B minified over this PR's base
• app: compiled hydrating (the same JSX todo app through hydrate(), compiled hydratable): over brotli cap by 179 B; minified 100,647 B vs 101,044 B recorded with the cap (−397 B) — 417 B of the 20 B minified allowance left; +31 B minified over this PR's base
• page: base server components (hydrating + dynamic + frames + sf reference): over brotli cap by 293 B; minified 101,502 B vs 101,559 B recorded with the cap (−57 B) — 77 B of the 20 B minified allowance left; +25 B minified over this PR's base
• page: live server components (base + live/GET + action + isPending/latest): over brotli cap by 295 B; minified 115,230 B vs 115,217 B recorded with the cap (+13 B) — 7 B of the 20 B minified allowance left; +27 B minified over this PR's base
• page: compiled base server components (the base page as JSX: templates with class/style/attributes/events, For/Show; no spread): over brotli cap by 169 B; minified 107,866 B vs 107,918 B recorded with the cap (−52 B) — 72 B of the 20 B minified allowance left; +19 B minified over this PR's base
• page: base + router (base page + @solidjs/router: createRouter, two routes, preload, useNavigate): over brotli cap by 392 B; minified 125,327 B vs 125,387 B recorded with the cap (−60 B) — 80 B of the 20 B minified allowance left; +22 B minified over this PR's base

❌ Fails the gate

• app: hydrating + every store primitive family: over brotli cap by 94 B; minified 97,070 B vs 97,045 B recorded with the cap (+25 B) — 5 B past the 20 B minified allowance; +25 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 341 B; minified 123,257 B vs 123,224 B recorded with the cap (+33 B) — 13 B past the 20 B minified allowance; +33 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 405 B; minified 140,518 B vs 140,485 B recorded with the cap (+33 B) — 13 B past the 20 B minified allowance; +33 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).}

[coveralls]

Coverage Report for CI Build 40801139067

Warning

Build has drifted: This PR's base is out of sync with its target branch, so coverage data may include unrelated changes.
Quick fix: rebase this PR. Learn more →

Coverage remained the same at 78.516%

Details

• Coverage remained the same as the base build.
• Patch coverage: No coverable lines changed in this PR.
• No coverage regressions found.

Uncovered Changes

No uncovered changes found.

Coverage Regressions

No coverage regressions found.

---

Coverage Stats

Relevant Lines:	1355
Covered Lines:	1125
Line Coverage:	83.03%
Relevant Branches:	1084
Covered Branches:	790
Branch Coverage:	72.88%
Branches in Coverage %:	Yes
Coverage Strength:	30.05 hits per line

---

💛 - Coveralls

[codspeed]

Merging this PR will degrade performance by 12.08%

1 warning

⚠️ 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

❌ 1 regressed benchmark
✅ 199 untouched benchmarks
⏩ 5 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
❌	projection derive: write one NESTED field (reference)	2.1 ms	2.4 ms	-12.08%

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-staged-recover (da52352) with next (56eb192)}

Footnotes

1. 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. ↩

[ryansolid]

CodSpeed projection derive: write one NESTED field (reference) at −12.35% on 9b777de is noise, not a regression from this PR:

• This benchmark swings ±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.
• A 200 B comment on next reproduced the same jump under callgrind.
• It flipped the same way on fix(signals): latest() opens no transaction for its verdict; its reader watches the seam (#4365) #4393, then the other way on fix(signals): a frame node re-run outside its hold re-derives the readers of the staging it replaces (#4378) #4390 (+12.42%).
• This PR doesn't touch the store engine.

---

Generated by Claude Code

[ryansolid]

The size check fails on purpose, and this PR won't change it. It goes over the recorded minified sizes in the scripts/size/scenarios.js ledger: hydrating + every store family +25 B, and compiled live server components +33 B. No frozen floor cap is raised.

The maintainer has asked for open PRs to merge as a batch without re-recording the ledger in each one. The ledger will be re-recorded once on next after the batch lands. Hello world is 9.99 KB brotli on this head.

---

Generated by Claude Code

[ryansolid]

CodSpeed flags the store projection-root-write "write one NESTED field (reference)" bench (−12%) on this PR, and on e36a6b9 also non-ascii-template "arrow in template". Both are known noise.

• projection-root-write: it swings ±12–15% 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 reproduces the jump (fix(signals): latest() of a memo born in a lane throws NotReady, not undefined (#4362, #4363, #4376) #4383). It flags fix(web): a hydrating hole holding a raw string re-claims its region (#4374) #4382 too, which changes only @solidjs/web.
• non-ascii-template: it's a server renderToString bench, and its base on next recorded a fast outlier.

---

Generated by Claude Code