feat(signals,solid,web,diagnostics): OBSERVE.records — one records channel on both platforms; the call record and its origin

[ryansolid]

Sentry-integration plan C4 (client half) and the channel unification it forced, plus J.28. Follows #3452 (C3) and #3453 (P4).

One channel: OBSERVE.records

The server records ("boundary", "invocation", "frame") lived on OBSERVE.server.records; the client had a separate, inherited observeServerFunctionCalls. Adding the client half of frames would have meant a third surface. Instead every record now rides one channel owned by the core:

• OBSERVE.records — subscribe(type, listener), observed(type), emit(type, event, live). One object per process, registered on globalThis under Symbol.for("@solidjs/signals/observe/records"), so a second copy of the core (a bundled server build instrumented through --import, SHAPE-NOTES §J.24) and wire layers bundled without a framework import (the server-function client) reach the same listener sets. Listeners snapshotted per emit; a throwing listener is reported and the rest run. Folds out of prod.
• Type layering: the core declares two empty interfaces, one augmenter each — RecordTypes (solid-js adds "boundary") extends HostRecordTypes (@solidjs/web adds "invocation", "frame", "call" through the solid-js re-export). A consumer reading OBSERVE.records off solid-js sees the whole catalogue; pinned by a type test against the built declarations.
• OBSERVE.server keeps only the trace slot.

The "frame" record, both sides

FrameEvent is FrameProducedEvent | FrameAppliedEvent, discriminated by side, the same census on both; the client half (applyFrameResponse, one per stream in a response) adds address (the as remap) and outcome: "truncated". The two join by id and version; the wire is the difference. Server census fix: regions counted a chunk type that does not exist — it now counts html chunks addressed to a child frame id — and shellMs is set by the stream's own shell only.

The "call" record, and its origin

One per server-function call made from the browser, at the caller's settle: { id, at, durationMs, method, outcome, status?, origin?, deferred? } with { args, response?, result? | error? } beside it. Joins the server's "invocation" by id. observeServerFunctionCalls is removed from both entries (no consumers in-repo, in the spike, in solid-router or solid-start; it also had to await the response body to show it to observers).

origin closes J.28: read at dispatch through the new OBSERVE.attribution.currentOrigin() — the provenance a root write made now would be stamped with: the interaction whose handler is running, the open navigation/effect/action frame, or inside a recompute the origin of the change that caused the run (a createAsync calling the server on a navigation's write is the navigation's, under its click). It is the engine's own ChangeOrigin object, so an observer files the call under InteractionEvent.origin / NavigationEvent.origin by identity, not by a time window. The installed hooks are also registered on globalThis (Symbol.for("@solidjs/signals/observe/attribution")) — the import-free reach the server-function client needs.

Tiers

The server-functions and frames client entries gain observe and development builds and export conditions. The server-functions client is now built with its flags replaced in every tier — before, _SOLID_DEV_ there was an unreplaced truthy string, so the dev cookie checks ran in prod. Dist scans pin the emitters (and the two registered names) out of the prod artifacts and into the observe/dev ones.

@solidjs/diagnostics v6

artifact.server → artifact.records: { boundary, invocation, frame, call }, always present, captured on both platforms including the browser bridge. JSONL: one line per record with type; recordCounts in the meta line.

Verified against real Sentry

The solid-sentry-spike (out of repo) was re-pointed to the channel on both platforms. Its instrument module now imports solid-js only — J.23 resolved — and the round trip joins: GET / → http.server with function.solid.direct and solid.boundary children sharing the boundary id; a click in the page → browser function.solid.call (status 200/500, sanitized error, origin: click on button#echo "echo(1)" through the compiled event binding) and server function.solid.rpc (unsanitized) in the same trace under the same id. The spike adapter files calls under the open interaction's transaction by identity.

Docs

RFC 08 (channel, all four records, currentOrigin), RFC 10, 11, 12; diagnostics README and skills; changeset.

[changeset-bot]

🦋 Changeset detected

Latest commit: 5680a39

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
solid-js	Patch
@solidjs/web	Patch
@solidjs/diagnostics	Patch
test-integration	Patch
@solidjs/element	Patch
@solidjs/h	Patch
@solidjs/html	Patch
@solidjs/universal	Patch
@solidjs/babel-plugin	Patch
@solidjs/compiler	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 35014116783

Coverage remained the same at 71.619%

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:	1014
Covered Lines:	772
Line Coverage:	76.13%
Relevant Branches:	790
Covered Branches:	520
Branch Coverage:	65.82%
Branches in Coverage %:	Yes
Coverage Strength:	14.99 hits per line

---

💛 - Coveralls

[codspeed]

Merging this PR will degrade performance by 2.42%

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 1 improved benchmark
❌ 1 regressed benchmark
✅ 173 untouched benchmarks

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
❌	projection derive: write one NESTED field (reference)	2.5 ms	2.8 ms	-10.37%
⚡	createStore setter: delete + set one root key (#3044 overlay)	582.6 µs	548.4 µs	+6.23%

Tip

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

---

_{Comparing server-frame-records (5680a39) with next (d2a36f5)}