fix(signals): a projection's reader pulls the derive it reads through (#4782) by ryansolid · Pull Request #4796 · solidjs/solid

ryansolid · 2026-11-12T09:15:26Z

What

A projection B whose derive reads the leaves of another projection A could run before A re-derived in the same flush. This happens when A's derive sits higher in the heap, for example because it reads a memo. B computed over A's previous answer, A re-derived, and B ran again. A projection R over N such Bs re-ran for each late A, so the flush was O(N²). The reporter's 10,000-node mount went from 2.3 s on rc.13 to 17.5 s on rc.21. This regressed in #3774.

Fix

pullFamily's settled branch read the derive untracked. An untracked read only wakes a lazy or dormant node (prepareComputed(el, false)). It does not refresh a stale one, and a store leaf has no _fn, so its reader never took a height from the derive. That branch now makes the same pull a tracked read of a memo makes (pullComputed(fw, reader)). The derive is brought up to date before any leaf is served, and the reader is lifted above it, so later flushes order it after the derive. It does this without linking, as before.

The pull is ordering only. The F-6 posture handoff and the #4491 stale-reader mark that follow a tracked read are untouched.

This is the reporter's experiment, adopted after checking it against the rules.

Rules checked

F-1: no derivation runs behind its input in a frame. B read A behind its input, which is the bug.
D-30: a derived store reads like its memo. A tracked read of a memo pulls it, and now a read through the family does too.
[2.0 / next] A reader that mounts while an unread derived store's flight is up shows the old question's answer (D-30, F-2) #4491 / D-25: the settled branch's verdict-reader mark is unchanged and still runs after the pull.

Size (local, brotli, against next @ `dc33de6`)

Core floor, hello world, simple app: 0
+ createStore: +12 B
Compiled scenarios were not measured locally (no native compiler in this sandbox). The CI size job covers them.

Tests

New packages/signals/tests/projection-reads-projection-4782.test.ts: the issue's shape at N=50, with the derive reading a memo and reading store keys only. Each B runs once and R once. On next the memo case gives 98 B runs and 49 R runs.
signals: 6,826 pass, 13 expected fail.

Generated by Claude Code

…#4782) pullFamily's settled branch read the derive untracked, which wakes a lazy node but does not refresh a stale one, and a leaf carries no height from its derive. A projection whose derive sits higher (it reads a memo) was read stale by a lower projection, which then ran again: O(N^2) per flush since #3774. The pull now makes the pull a tracked read makes: the derive brought up to date, the reader lifted above it. Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_0117BoWNSSgtemsSYsaX2hXE

changeset-bot · 2026-11-12T09:15:31Z

🦋 Changeset detected

Latest commit: 7817955

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 · 2026-11-12T09:18:37Z

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)	7.44 KB	0 B	0 B	−84 B	7.43 KB	⚠️ over by 11 B, 104 B minified headroom
signals: + createStore	14.91 KB	+12 B (+0.1%)	+36 B	−588 B	15.13 KB	✅
signals: + isPending/latest	12.22 KB	0 B	0 B	−266 B	12.27 KB	✅
signals: + until/refresh (truth readers, no optimism)	8.10 KB	0 B	0 B	−11 B	8.12 KB	✅
app: render + one signal (the simple-app floor)	9.73 KB	0 B	0 B	−399 B	9.81 KB	✅
app: hydrating (no stores) with Show/For/Loading/Errored/lazy	18.96 KB	0 B	0 B	−367 B	19.11 KB	✅	lazy-page.js 0.04 KB
app: hydrating + every store primitive family	34.26 KB	−22 B (−0.1%)	+36 B	−518 B	34.36 KB	✅	lazy-page.js 0.04 KB
app: CSR with Show/For/Loading/Errored/lazy	13.72 KB	0 B	0 B	−367 B	13.77 KB	✅	lazy-page.js 0.04 KB
app: CSR, observe tier (same app on the `observe` artifacts)	15.41 KB	0 B	0 B	−318 B	15.50 KB	✅	lazy-page.js 0.04 KB
app: CSR, observe tier + attribution engine enabled	30.04 KB	0 B	0 B	−292 B	30.16 KB	✅	lazy-page.js 0.04 KB
app: compiled floor (one template, one text hole, one delegated click)	9.93 KB	0 B	0 B	−452 B	9.99 KB	✅
app: compiled CSR (JSX todo app: spread/merge/omit, events, class/style, keyed For, Show, Loading + lazy, store)	26.33 KB	−29 B (−0.1%)	+36 B	−1105 B	26.75 KB	✅	stats.js 0.18 KB
app: compiled hydrating (the same JSX todo app through hydrate(), compiled hydratable)	32.66 KB	+65 B (+0.2%)	+36 B	−1100 B	33.00 KB	✅	stats.js 0.20 KB
frames: eager client consumer (frames client + transport, lazy codec)	11.75 KB	0 B	0 B	0 B	11.77 KB	✅
page: base server components (hydrating + dynamic + frames + sf reference)	33.96 KB	−65 B (−0.2%)	+9 B	−268 B	34.11 KB	✅	assets.js 0.78 KB, decode.js 7.09 KB, lazy-page.js 0.04 KB, regions.js 1.06 KB, trace.js 9.06 KB, wire.js 3.48 KB
page: live server components (base + live/GET + action + isPending/latest)	40.30 KB	−17 B (−0.0%)	+15 B	−667 B	40.43 KB	✅	assets.js 0.78 KB, decode.js 7.09 KB, lazy-page.js 0.04 KB, regions.js 1.06 KB, trace.js 9.05 KB, wire.js 3.48 KB
page: compiled base server components (the base page as JSX: templates with class/style/attributes/events, For/Show; no spread)	35.94 KB	+71 B (+0.2%)	+3 B	−560 B	36.06 KB	✅	assets.js 0.78 KB, decode.js 7.09 KB, regions.js 1.06 KB, sc-comments.js 0.20 KB, trace.js 9.07 KB, wire.js 3.48 KB
page: compiled live server components (the compiled base page + live/GET + action + isPending/latest)	44.27 KB	0 B	0 B	−680 B	44.39 KB	✅	eager (counted): web.js 23.64 KB; assets.js 0.78 KB, decode.js 7.09 KB, regions.js 1.08 KB, sc-comments.js 0.19 KB, trace.js 9.07 KB, wire.js 3.48 KB
page: base + router (base page + @solidjs/router: createRouter, two routes, preload, useNavigate)	41.51 KB	−64 B (−0.2%)	+9 B	−355 B	41.60 KB	✅	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 3.59 KB, trace.js 9.02 KB, wire.js 3.48 KB
page: live + router (live page + @solidjs/router: createRouter, two routes, preload, useNavigate)	50.04 KB	0 B	0 B	−676 B	50.20 KB	✅	eager (counted): client.js 28.91 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 3.31 KB, trace.js 9.01 KB, wire.js 3.50 KB
server: floor (getRequestEvent + isServer)	1.27 KB	0 B	0 B	0 B	1.29 KB	✅
server: renderToString (the server-render floor)	22.93 KB	0 B	0 B	+2 B	22.92 KB	⚠️ over by 9 B, 18 B minified headroom

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

signals: core floor (createSignal/Memo/Effect/Root/flush): over brotli cap by 11 B; minified 19,839 B vs 19,923 B recorded with the cap (−84 B) — 104 B of the 20 B minified allowance left; +0 B minified over this PR's base
server: renderToString (the server-render floor): over brotli cap by 9 B; minified 77,890 B vs 77,888 B recorded with the cap (+2 B) — 18 B of the 20 B minified allowance left; +0 B minified over this PR's base

_{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 · 2026-11-12T09:23:05Z

Coverage Report for CI Build 43189861931

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 78.945%

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:	1372
Covered Lines:	1143
Line Coverage:	83.31%
Relevant Branches:	1112
Covered Branches:	818
Branch Coverage:	73.56%
Branches in Coverage %:	Yes
Coverage Strength:	30.23 hits per line

💛 - Coveralls

codspeed · 2026-11-12T09:26:44Z

Merging this PR will degrade performance by 5.19%

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

⚠️ 1 benchmark with a significant change ran on different hardware

Its base and head results were measured on machines with a different CPU, core count or memory, so the change might not come from the code. Run base and head on the same runner type.

Learn more about Macro Runners

See affected benchmarks

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

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	`BASE`	`HEAD`	Efficiency
❌	`hn story page (1,475 anchors): no link handler`	23.4 ms	24.7 ms	-5.19%

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-c5xes4-4782 (7817955) with next (0895916)}

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 merged commit 6681c1b into next Nov 12, 2026
13 of 14 checks passed

ryansolid mentioned this pull request Nov 12, 2026

[2.0 / next] mapArray over a keyed projection maps an empty slot after a declined removal guess (regression from #4785) #4821

Closed

Navigation Menu

Repository navigation

fix(signals): a projection's reader pulls the derive it reads through (#4782) - #4796

fix(signals): a projection's reader pulls the derive it reads through (#4782)#4796
ryansolid merged 1 commit into
nextfrom
claude/project-thread-c5xes4-4782

ryansolid commented Nov 12, 2026

Uh oh!

changeset-bot Bot commented Nov 12, 2026

Uh oh!

github-actions Bot commented Nov 12, 2026

Uh oh!

coveralls commented Nov 12, 2026

Uh oh!

codspeed Bot commented Nov 12, 2026

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

ryansolid commented Nov 12, 2026

What

Fix

Rules checked

Size (local, brotli, against next @ dc33de6)

Tests

Uh oh!

changeset-bot Bot commented Nov 12, 2026

🦋 Changeset detected

Uh oh!

github-actions Bot commented Nov 12, 2026

Size (brotli, eager entry chunk)

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

Uh oh!

coveralls commented Nov 12, 2026

Coverage Report for CI Build 43189861931

Coverage remained the same at 78.945%

Details

Uncovered Changes

Coverage Regressions

Coverage Stats

💛 - Coveralls

Uh oh!

codspeed Bot commented Nov 12, 2026

Merging this PR will degrade performance by 5.19%

Performance Changes

Footnotes

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Size (local, brotli, against next @ `dc33de6`)