feat(signals,solid): pay-for-use tree-shaking phase 3 — audited cleanups (#2883)

[ryansolid]

Phase 3 of #2883 — every item here passed the cost/benefit audit (wins-population / pays-population / complexity, adversarially verified with byte-exact reproduction). Two analyst-proposed items were rejected by ruling: the event-delegation seam (delegation is the blessed path — all apps are assumed to have it) and the snapshot-capture seam (win too small for a new seam under the weight gate).

Measured (esbuild, minify, --mangle-props='^_', gzip -9)

Fixture	before	after	Δ gzip
Minimal app (published dist)	27,570 / 10,870	26,429 / 10,432	−438
CSR app with <Loading>/lazy	35,565 / 13,935	32,067 / 12,601	−1,334*
Signals core floor	18,921 / 7,687	18,539 / 7,515	−172
Full signals bundle	55,366 / 20,587	54,587 / 20,315	−272 (hook-tax clawback)

*includes ryansolid/dom-expressions#543 (DOMElements dedup) propagating through the rebuilt web runtime.

Cumulative over phases 1–3: minimal app 12,656 → 10,432 gz (−17.6%), signals floor 8,885 → 7,515 gz (−15.4%).

Changes

@solidjs/signals

• Effect re-enqueue block (4 copies) and the zombie/dirty queue ternaries dedupe into shared enqueueSub/queueFor — the hot write-propagation loop was microbenched against baseline (interleaved 100-fanout write+flush ×20k), no regression.
• Boundary/reveal internal method names _-prefixed so property mangling reaches them (verified not part of any exported surface; the server's revealGroup.register is a different object).
• Prod error strings trimmed to their diagnostic codes; dev builds keep the full sentences.
• Build pipeline fix: the prod dist was shipping with zero /*@__PURE__*/ annotations — rollup-plugin-prettier strips them, and terser only re-emits them with preserve_annotations. Prettier is off the prod tree (the mangle post-pass beautifies anyway), terser preserves, and a new scripts/check-pure.mjs guard fails the build if src has annotations and dist has none.

solid-js (client)

• sharedConfig.getNextContextId was defined in the object literal, retaining getContext/NoHydrateContext/getNextChildId and the signals context+error machinery in every CSR bundle; it now installs from enableHydration() (type becomes optional; both callers sit behind sharedConfig.hydrating).
• lazy()'s hydration-module lookup (positional id + _$HY.modules + error string) moves behind an _lazyHydrationLookup slot installed by enableHydration() — the 12th instance of the existing override-slot pattern.
• MockPromise's static {} block marked the class side-effectful, defeating DCE in every client bundle (the code comment claiming it shakes was wrong); it's now a PURE-annotated factory that shakes with its only consumer.

Packaging

• unpkg/jsdelivr fields pointed at CJS files (browser-fatal on bare CDN URLs — verified over the network); they now point at browser production ESM for all three packages.

Verification

984 signals + 456 solid + 501 solid-web tests green; both typechecks clean; check-pure guard active in the signals build. Rejected-by-audit items are documented in the analysis (dist minification, standalone CDN bundle, reconcileArrays detach, hydration guards in insert, flatten split, optimistic helper dedup — each with the measured reason).

🤖 Generated with Claude Code

[changeset-bot]

🦋 Changeset detected

Latest commit: 5946c0e

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

This PR includes changesets to release 9 packages

Name	Type
@solidjs/signals	Patch
solid-js	Patch
test-integration	Patch
@solidjs/element	Patch
@solidjs/h	Patch
@solidjs/html	Patch
@solidjs/universal	Patch
@solidjs/web	Patch
babel-preset-solid	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

[cursor]

Bugbot is not enabled for your account, so this pull request was not reviewed.

Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs.

[coveralls]

Coverage Report for CI Build 29448338455

Warning

No base build found for commit d94d5c3 on next.
Coverage changes can't be calculated without a base build.
If a base build is processing, this comment will update automatically when it completes.

Coverage: 74.74%

Details

• Patch coverage: 7 uncovered changes across 1 file (15 of 22 lines covered, 68.18%).

Uncovered Changes

File	Changed	Covered	%
packages/solid/src/client/hydration.ts	20	13	65.0%
Total (2 files)	22	15	68.18%

Coverage Regressions

Requires a base build to compare against. How to fix this →

---

Coverage Stats

Relevant Lines:	596
Covered Lines:	477
Line Coverage:	80.03%
Relevant Branches:	461
Covered Branches:	313
Branch Coverage:	67.9%
Branches in Coverage %:	Yes
Coverage Strength:	14.28 hits per line

---

💛 - Coveralls

[codspeed-hq]

Merging this PR will not alter performance

✅ 7 untouched benchmarks
⏩ 113 skipped benchmarks¹

---

_{Comparing size/2883-phase-3 (5946c0e) with next (d94d5c3)}

Footnotes

1. 113 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. ↩