fix(signals): optimistic lanes release when nothing authoritative remains; shared render effects don't entangle unrelated writes (#3426, #3427, #3407)

[ryansolid]

Fixes #3426, fixes #3427, fixes #3407.

One theme: the pure graph does not wait where it is not forced to, and a hold is a fact about what a pass observed — never about which reader happened to exist.

#3426 — the last async reader unmounting mid-action releases the frame

const details = createMemo(() => delay(1500, value()));   // derived from the optimistic value
run = action(function* () { setValue(1); yield delay(3000); });
<Show when={show()}><Details /></Show>                    // unmounted at 2500

The lane held Value: 1 on the Details reader observing details pending. A live action parks its transaction without a settle verdict, so that reader's registration was never pruned when it died; the lane waited for a flight nobody observed.

Rule: the hold check prunes as it reads. reporterBlocksSource is now sourceObserved — the live-reporter test shared by the settle verdict (transitionComplete), the lane hold (waitingTransition) and the landing. The frame reveals at the unmount and reverts when the action ends, like a plain write would.

#3427 — the action body ending starts the correction

const details = createMemo(() => delay(1000, value()));
run = action(function* () { setValue(1); yield delay(500); });

The settle waited for details(1) to land (3000: Value: 1 | Details: 1 flashed), then reverted and re-asked details(0) (4000) — a waterfall with a flash, for a guess the action had stopped justifying at 2500.

Rule (A18 body-end corollary): once the bodies have ended and nothing authoritative is in flight — no override node's own source (transitionBlocked), no held flight that does not derive from an override — each override's truth supersedes it, exactly as an arriving differing truth does (#3331): the graph re-derives from the truth as the transaction's held work and settles when that lands (3500, nothing visible changes). Display is untouched: the override stays on screen until commit. Carve-outs, all existing rulings: a co-written pending flag stays through the action's own plain load (setSaving(true); setPage(2) — the attribution-holds saving-indicator pin depends on this); optimistic store edits keep the settle-then-revert order; companions snap at settlement. The typical refresh-inside-the-action pattern never hits the hook — the refresh is the transaction's own authoritative work and holds through.

Mechanism: Transition._acted (set by action, merged), GlobalQueue._endOptimism called from flush after the heap and before the verdict; a supersession re-runs the heap in the same pass.

#3407 — a shared render effect no longer entangles an unrelated sync write

text(() => `${b()}:${detailsA()}`);     // one hole, two bindings
bump = action(function* () { setB(p => p + 1); });

A plain setB published 1:0 beside the pending detailsA; the same write inside an action merged into the async's transaction and waited (0:0 → 2:1). recompute's head re-entered a stamped effect's transaction whenever any other transaction was active — before the pass ran, before anything could have been observed.

Rule (A15 shared-hole corollary): a render effect's pass belongs to whatever dirtied it. The stamp re-entry is memo-only (a memo's value is its transaction's work); a stamped effect dirtied by another writer runs that writer's pass, reads the held flight as a stale reader (committed value, coherent with the flight's committed inputs) and publishes with the writer. Only a pass that actually observes a pending flight — the reveal carve-out refused — joins it. The re-entry's one legitimate job, completing every reveal that discovered a flight, moves to the landing: settleTransition → enterWaiting folds in every transaction still waiting on the node. Consequence beyond the issue: two independent flights read in one hole now land at their own times (next: one unit).

Verification

• @solidjs/signals: 169 files / 1794 tests green. No existing expectation moved; the A15 reveal-completion pins (spec-async-semantics, reveal-carve-out, stale-read-uninitialized-cross-transition) are the regression net for the landing fold.
• New pins: tests/optimistic-lane-release.test.ts (2.0.0-rc.8 optimistic update keeps waiting after its last async reader unmounts #3426, 2.0.0-rc.8 optimistic update first reveals after its action body has finished #3427 ×2, co-written flag), tests/shared-effect-no-entangle.test.ts (2.0.0-rc.8 shared render effect entangles a synchronous action with an unrelated pending update #3407 plain vs action, carve-out plain vs action).
• Core floor 23,058 → 23,195 (+137 B: +78 2.0.0-rc.8 optimistic update keeps waiting after its last async reader unmounts #3426/2.0.0-rc.8 optimistic update first reveals after its action body has finished #3427, +59 2.0.0-rc.8 shared render effect entangles a synchronous action with an unrelated pending update #3407, after a golf pass), noted in treeshake.test.ts; six brotli caps ratcheted with notes in scripts/size/.size-limit.js.
• Docs: SPEC-ASYNC-SEMANTICS.md A15 amended in place (shared-hole corollary), A18 body-end corollary; INTERNALS-ASYNC-STATE.md §2/§3 mechanism notes; RULES-INDEX.md regenerated (A15 → amended). Changeset included.

Based on next after #3432.

Co-authored-by: Claude via Cursor noreply@cursor.com

[changeset-bot]

🦋 Changeset detected

Latest commit: 9e21d80

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

This PR includes changesets to release 11 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

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

[coveralls]

Coverage Report for CI Build 34885888106

Coverage remained the same at 71.842%

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:	1007
Covered Lines:	772
Line Coverage:	76.66%
Relevant Branches:	790
Covered Branches:	519
Branch Coverage:	65.7%
Branches in Coverage %:	Yes
Coverage Strength:	15.03 hits per line

---

💛 - Coveralls

[codspeed-hq]

Merging this PR will not alter performance

✅ 160 untouched benchmarks

---

_{Comparing fix/optimistic-lane-release (9e21d80) with next (a031271)}