@ultimat3/realtime 21.0.0 → 22.0.0

package/CLAUDE.md

@@ -6,991 +6,308 @@ Tier 3 package. Channels, live queries, local-first sync. One protocol for all t
 
 | May import | Must not |
 |---|---|
-| `@ultimat3/core`, `@ultimat3/query`, `@ultimat3/entity` (`/record` only: `recordTypeForTable`, server-side) | anything tier 4+ (`render`, `pwa`, `mcp`, `ui`, `cli`) |
+| `@ultimat3/core`, `@ultimat3/query`, `@ultimat3/entity` (`/record` only, server-side) | anything tier 4+ (`render`, `pwa`, `mcp`, `ui`, `cli`) |
 | `@ultimat3/policy` **only via** `@ultimat3/query`'s `guard` | a second authz path of any kind |
 | — | `solid-js` (each island bundle installs its own signal factory: `installRealtime`) |
-| `nats` (the one external dependency, pinned exact) — from `nats-lib-client.ts`, and no other file | `nats` from anywhere else: a second importer is the failure this row exists to prevent. Every other file is written against the port in `nats-client.ts` |
+| `nats` (the one external dependency, pinned exact) — from `nats-lib-client.ts` and no other file | `nats` from anywhere else. Every other file is written against the port in `nats-client.ts` |
 
-## Rules
+## Entries and bundling
 
-- **Two entries, and a name lives in exactly ONE of them (2026-08-22, BREAKING).** `.` is the
-  client half — `use-*` hooks, `page-*`, `reactivity`, `record-*`, `client*`, `browser-socket`,
-  `live-rows`, `apply-patches`, `offline-queue`, `sync-protocol`, `json`, `cursor`, `errors`, and
-  the client's half of `thundering-herd`. `./server` (`src/server.ts`) is everything that touches `nats`, Postgres, the
-  sync node, the channel hub, the live-query registry or the fanout. The reason is measured, not
-  aesthetic: `nats` `require()`s `stream/web`, so one barrel carrying `openNatsClient` beside
-  `useLive` failed `bun build --target=browser` with
-  `Browser build cannot require() Node.js builtin: "stream/web"` — the island `wiki/Realtime.md`
-  promises could not be built at all. Two build errors hold it: `packages/cli/src/realtime-browser-barrel.test.ts`
-  bundles a `useLive`-only entry for the browser, and `barrel-split.test.ts` refuses a name
-  exported from both (values through the namespace objects, types off the source text, because a
-  type-only export leaves no runtime entry). `errors.ts` is deliberately whole on `.`: every code
-  reaches the wire through `toWireError`, so a client must be able to name any of them, and the
-  module is already in the client graph via `sync-protocol`.
-- **`errors.ts` is the code TABLE plus the client-reachable refusals; two neighbours hold the rest,
-  and every name is still exported from `./errors`** (2026-08-23, at the 500-line ceiling).
-  `realtime-error.ts` holds the base class alone and `replication-errors.ts` the four Postgres ones
-  — the only codes no browser can reach. The base needs its own module rather than a re-export:
-  `extends` runs at module evaluation and imports hoist above it, so a `replication-errors` that
-  imported the base back out of `errors.ts` would read it in its temporal dead zone. Neither
-  neighbour runs anything at import, which is what keeps `sideEffects` naming `errors.ts` alone
-  true — `registerErrorCodes()` stays there, unconditional, and `bun run side-effects` is the check.
-- **`sideEffects` is the ARRAY `["./src/errors.ts"]`, never `false` and never absent.** Absent was
-  what made the failure above unrecoverable — with no field a bundler must assume every module has
-  effects, so nothing was tree-shaken and `nats` came along with `useLive`. Measured, not guessed:
-  `bun run side-effects --explain --json` prints what the tree actually does at import time.
-  `false` would drop `registerErrorCodes()` and every `REALTIME_ERROR_TITLES` entry with it.
-  **The array alone would have fixed the build** and is not why the split exists: tree-shaking is
-  a bundler's discretion, `export * from` or a namespace import defeats it, and "the client entry
-  cannot reach the bus" is a contract rather than an optimisation.
-- **Two entries means a specifier naming a third does not resolve, and a `fix:` is pasted.**
-  `local-store.ts`'s `X_NOT_IMPLEMENTED` told the caller to import `createOpfsLocalStore` from
-  `@ultimat3/realtime/browser` — a subpath `exports` has never declared — so the one instruction
-  the refusal carried ended in a module-resolution failure, in the package whose own rules cite
-  axiom 4. Its alternative, `persist: false` on the query, was the same defect twice: `query()`
-  does not accept `persist` either. `fix-specifier.test.ts` is the build error — every
-  `@ultimat3/realtime/<subpath>` written in shipped source must be a key of `exports`, comments
-  included, because a comment naming a subpath that does not exist is the next fix line's source.
-  The OPFS refusal it was written for is deleted with `local-store.ts` (21.0.0).
-- **`@ultimat3/realtime/server` needs its own `paths` entry in `tsconfig.base.json`**, beside
-  `@ultimat3/admin/dev`'s. `@ultimat3/*` maps `realtime/server` to `packages/realtime/server/src`,
-  which does not exist, and the root program has no `node_modules/@ultimat3` symlink to fall back
-  through — so `scripts/**` (which the root `tsc -b` compiles) reports `TS2307` without it. The
-  workspace packages resolve through their own `node_modules` and never needed it, which is what
-  makes the failure look local to `scripts/`.
+- **Two entries; a name lives in exactly ONE.** `.` is the client half (`use-*`, `page-*`,
+  `reactivity`, `record-*`, `client*`, `browser-socket`, `live-rows`, `apply-patches`,
+  `offline-queue`, `sync-protocol`, `json`, `cursor`, `errors`, the client half of
+  `thundering-herd`). `./server` (`server.ts`) is everything touching `nats`, Postgres, the sync node,
+  the hub, the registry or the fanout (`nats` requires `stream/web`, which a browser build cannot
+  link). `packages/cli/src/realtime-browser-barrel.test.ts` bundles a `useLive`-only entry;
+  `barrel-split.test.ts` refuses a name exported from both. `./boot` and `./sync-worker` are the page
+  boot and the SharedWorker entry.
+- **`errors.ts` is the code table plus the client-reachable refusals**; `realtime-error.ts` holds
+  the base class alone (an `extends` must not read it in a TDZ) and `replication-errors.ts` the
+  Postgres ones. Every name is still exported from `./errors`. A browser path loads `page-errors.ts`,
+  never the table (`page-errors-bundle.test.ts`).
+- **`sideEffects` is the ARRAY `["./src/errors.ts"]`**, never `false` (drops `registerErrorCodes()`)
+  and never absent. `bun run side-effects` is the check.
+- **Every `@ultimat3/realtime/<subpath>` written in shipped source must be a key of `exports`**,
+  comments included — `fix-specifier.test.ts`.
+- **`@ultimat3/realtime/server` needs its own `paths` entry in `tsconfig.base.json`** (the wildcard
+  maps it to a directory that does not exist; `scripts/**` reports `TS2307` without it).
+- A browser file imports core from `@ultimat3/core/page` (`packages/core/src/page-bundle.test.ts`,
+  `packages/query/src/client-bundle.test.ts`).
 
-- Policy is evaluated **once per subscriber**, never once per query. `live-query.test.ts` proves it
-  for a hand-written definition and `live-definition.test.ts` proves it for a real declared
-  `query({ live: true })` — the second one matters, because a rule that only holds for test fakes
-  is a rule no declaration can reach.
-- **Every numeric option is refused when it is not a FINITE number, `As of 2026-08-26`.**
-  `@ultimat3/core`'s `finite-option.ts` holds the one refusal, `finiteOption()`. `??` guards
-  nullish and `NaN` is not, so `Number(process.env.X)` on an unset variable reaches the bound
-  intact, and `Math.max`/`Math.min`/`Math.floor` propagate rather than validate — `AcceptBudget`
-  was `Math.max(1, options.perSecond)` and admitted every accept, because `NaN < 1` is false.
-- **"Pinned at zero" is a claim about the RATCHET, not about this package, and the two came apart
-  once (`As of 2026-08-26`).** `bun run finite-bounds` is a count of options defaulted with `??`
-  and never screened; `realtime` is absent from `scripts/lib/finite-bounds-pins.ts`, which is that
-  count reading zero. It read zero while **two** options were unscreened, because the rule's own
-  header names the shape it cannot see: an option with **no `??` default**. Both had one spelling
-  — a value the caller either supplies or does not, forwarded or compared as-is.
-  `SubscriptionBook`'s `maxPerTenant` (`count >= NaN` is false, so the only cap spanning the
-  sockets of one tenant was OFF; measured, 5,000 subscribes admitted under `maxPerTenant: NaN`,
-  and it sat two lines under the screened `maxPerSocket`), and `openNatsClient`'s
-  `maxReconnectAttempts`, handed to the library unscreened — measured, the dial then never
-  returns. It also cannot see WHERE a screen runs, which is the bullet below. So: a package's
-  numeric options are audited by reading its option interfaces, and the ratchet is the floor under
-  that, never the proof of it. `SyncGrant.expiresAt` is deliberately NOT on that list and is the
-  one number left with the same shape: it is DATA an app's `authenticate` returns, not a
-  configuration option, and `expiresAt: NaN` makes `GrantBook.expired()` skip that grant forever —
-  the same outcome as omitting it, which is already a supported spelling.
-- **A ceiling is refused where the object is BUILT, never inside a callback the runtime invokes
-  per connection (`As of 2026-08-26`).** `maxBufferedBytes`, `maxDroppedFrames`,
-  `maxFramesPerSecond` and `frameBurst` were screened only in `SyncSocket`'s constructor, which
-  `sync-node` runs inside `websocket.open`, which Bun runs SYNCHRONOUSLY inside `server.upgrade`.
-  Measured: `createSyncNode` did not throw, `/healthz` and `/readyz` both answered, `ready` was
-  true, and every upgrade threw `X_INVARIANT` with the node holding zero sockets — a node that
-  boots green and refuses every client, whose cause is one frame inside the runtime. They are
-  `socketCeilings()` in `sync-node-bounds.ts` now, called once by `createSyncNode`, and the
-  per-socket screen STAYS beside it: `SyncSocket` is exported and an app may build one directly,
-  which is the layered repair `finite-bounds`' own header prescribes for a check in another file.
-  `undefined` stays `undefined` — `SyncSocket` owns those four defaults, and a second spelling of
-  one is a number that can drift from it.
-- **An upgrade that THROWS gives the grant back, not only one that answers `false`
-  (`As of 2026-08-26`).** `handleUpgrade` records the grant before `server.upgrade` because `open`
-  reads it, and released it on the `false` branch alone — so a throw out of `open` left one
-  `GrantBook` entry per connection ATTEMPT, unreapable: `sweepGrants` only visits a grant carrying
-  an `expiresAt`, which `authenticate: async () => ({ actor })` does not produce. Measured, 20
-  failing upgrades left 20 grants. The `try` around `server.upgrade` rethrows untouched — the throw
-  is the operator's diagnosis and that line owes it the release, not a verdict.
-- **Every ceiling is refused when it is not a FINITE number, `As of 2026-08-26`.**
-  `AcceptBudget`'s `perSecond`/`burst`, `SyncSocket`'s `maxBufferedBytes`/`maxDroppedFrames` and
-  `SocketRegistry`'s `idleTimeoutMs` throw `X_INVARIANT` at construction on `NaN` and `±Infinity`.
-  Every comparison against a non-finite bound is FALSE, so each guard did not loosen — it switched
-  off, silently. Measured: `tryAccept()` asks `tokens < 1`, so `perSecond: NaN` admitted every
-  accept, herd included, on the node path AND on the per-socket frame flood budget;
-  `maxBufferedBytes: NaN` made `send()` answer TRUE with 10 MB queued, so a discarded frame was
-  neither counted in `channel_frames_dropped_total` nor desync-marked — the delivery-accounting
-  guarantee this file states, voidable through one option; `idleTimeoutMs: NaN` left a socket idle
-  for 10,000,000 ms out of `idle()`. `??` guards only nullish and `Math.max` is a clamp, not a
-  validator: `Number(process.env.X)` on an unset variable reaches the comparison intact.
-- **A SQLSTATE off the wire is DATA, so `pg-wire.ts`'s `FIXES` is read with `Object.hasOwn`.**
-  `FIXES['constructor']` answered the `Object` function — not nullish, so `?? GENERIC_FIX` never
-  fired — and `UltimateError` then ran `singleLine(fn)`: a `TypeError` out of the constructor of
-  the error that exists to explain the failure, so the caller lost `X_REPLICATION_FAILED`, its
-  cause and its fix at once. This was `realtime`'s one `proto-index` pin, and the pin's stated
-  reason ("keyed by a replication op this package declares") described a different read.
-- **A name nothing registered is `X_LIVE_QUERY_UNKNOWN`, never `X_PROTOCOL_VERSION`.** The frame
-  parsed and the version matched — one string in it names nothing — so "x build && redeploy the
-  client" is the one instruction that cannot work: a rebuilt client spells the typo the same way,
-  and the registry that would have shown the mismatch never gets opened. The fix line is
-  `x queries list --json`. The name the client sent is echoed back; the registry is never
-  enumerated over the wire, because an unauthenticated socket walking `a`…`zz` is not entitled to
-  a list of every read this app declares. It is a client fault, so it never pages anyone. `fix` is
-  the command and nothing else — what to do with what it prints is in `cause`, because a fix line
-  is pasted into a shell and prose appended to one is a command that does not run.
-- **One build per `(query, input)`, and the window reads through it.** `target.live()` produces the
-  descriptor *and* runs the read (`LiveQuery.execute`) — a second subject-less `sourceFor` for the
-  rows was two descriptions of one read that agreed only by luck, at twice the parse and twice the
-  `sql()` per query id. Both halves must come from one build or the matcher patches a window it
-  never saw: `live-definition.test.ts` proves it with a declaration whose rows carry the number of
-  the build that produced them, and under the old code the subscriber was served build 2's rows.
-  `execute()` runs on every call rather than memoising — a client joining an existing subscription
-  sees the rows as they are now.
-- What `liveQueryDefinition` caches per query id is the compiled source, the shape, the matcher and
-  the shared row window. What it must never cache is a decision. It builds that shared half with
-  `enforce: false` **on purpose**: a source compiled under the first subscriber's authority and
-  then keyed by query id is that subscriber's entitlements becoming everyone's. `authorize` is
-  still the subscribe-time decision and still runs per socket.
-- Every policy call in `live-query.ts` takes a `Subscriber`. That is the enforcement: there is no
-  path through the gate that reads a query id and no actor.
-- **The row policy always sees the *whole* row from the shared window, never a partial patch — and
-  a window that does not hold the row is not a partial one.** An update patch carries the changed
-  columns plus the id, so merging it onto nothing and calling that a row hands `visible` a
-  `undefined` for every column the change did not touch: fail-closed for `row.ownerId === actor.id`,
-  and a leak for every `!row.private`. So a patch whose row the window does not hold is **withheld**
-  — dropped, or the one `delete` frame that tells a subscriber holding it that it is gone. It is
-  neither `rowsDenied` nor `gateFailures`: nothing decided anything, the window simply *is* the
-  result set. The one path that could meet an empty window is a delta resume onto an entry nothing
-  has read yet, and `subscribe` fills it first (`entry.lsn === ''`) rather than withholding
-  everything — conditional on purpose, because re-reading per resuming subscriber is exactly the
-  cost a delta resume exists to skip in a restart storm.
-- **A denial is a decision; everything else is a failure, and the two never share an answer.** A
-  bare `catch { return false }` in the row gate read a dead pool as "you may not see this row" —
-  the rows left the screen, `live.rows_denied` counted the drop, and the outage shipped as a
-  permission change. `visibleWithPolicy` matches `QueryDeniedError` (the only thing `guard` throws
-  for a decision) and rethrows the rest; `subscriber-gate.ts` and `reauthorize` ask
-  `isPolicyDenial(error)` instead, because `authorize` and `visible` are caller-supplied functions
-  and the answer has to come off the error's code. What a failure costs is decided per surface: a
-  snapshot **raises** out of `subscribe` (a short result set is indistinguishable from a correct
-  one), a delivery desyncs that **one** subscriber and lets the fanout finish, and a `reauthorize`
-  keeps the subscription — destroying it would report a timeout as a revoked grant, and a client
-  does not resubscribe to a denial. Every failure is counted as `gateFailures` and reported through
-  `onGateFailed`, never through `onRowDenied`: an alert fires on one of them.
-- **One serial lane per query id, and it is the only thing that orders a fanout.** Nothing upstream
-  does: `sync` fires `void registry.deliver(change)` straight off the bus subscription, so two
-  changes arriving back to back both start, both write `entry.rows`/`entry.lsn`, and both await
-  their way through a per-subscriber gate in between. Unordered, a subscriber is handed lsn 2 and
-  then asked to fold lsn 1 on top of it, its cursor rewound to 1 — a reconnect then replays what it
-  already applied, over newer state, and the row stays at the older value. `WindowLock` (`run`)
-  gives each entry a FIFO lane. `deliver` *enters* every lane before awaiting any of them, and no
-  fanout ever takes a second lane, so holding all of them at once cannot be a cycle — and two
-  deliveries queue onto each query id in call order, which is what makes "per query id, not per
-  node" true. Awaiting one entry before entering the next was two bugs in one line: one slow policy
-  pass set the whole node's pace, and a lane that threw ended the loop, so every entry behind it
-  missed the change with **nobody desynced** — the silent divergence `markDesynced` exists to
-  prevent. A lane that fails now desyncs its own subscribers and the first failure still reaches
-  the caller, but it costs one query id. The lane chains on a settled shadow of each task: one
-  fanout that threw must not reject every fanout behind it.
-- **Two reads of one entry are ordered by a READ GENERATION, never by an lsn.** `QueryEntry.lsn`
-  is optional — a definition with no lsn provider answers `''` from every snapshot — so the
-  never-backwards rule expressed purely in lsn terms read `'' >= ''` as "newer" and let the older
-  of two concurrent reads land on top of the newer one's window. The interleaving: a cold
-  subscriber issues P1; the change stream skips a sequence and `registry.invalidate()` marks the
-  entry; a second cold subscriber forces P2, which **clears `stale` on the way in**; P2 lands with
-  the post-gap rows; P1 lands last and overwrites them. `stale` is false, so `fanoutChange`'s
-  repair never fires, the next change patches the pre-gap window and re-snapshots every desynced
-  subscriber out of it — permanently stale on a healthy socket, which is the exact outcome `stale`
-  exists to prevent. `entry.generation` is bumped in `startRead` and `entry.applied` records the
-  newest read whose rows are on the window: an *identity* check, the same one `startRead` makes on
-  `entry.reading` one function down and `packages/cache/src/single-flight.ts:70` makes for the same
-  reason. The lsn guard stays beside it for the other question — a read that resolved behind a
-  *change* the fanout already folded — because those are two orderings and neither answers the
-  other.
-- **The definition's read is once per entry, not once per subscriber.** A cold subscriber arriving
-  while another's read is in flight joins that read — N cold subscribers on one query id being N
-  reads is the shared window not existing. It is a share, not a cache: the in-flight promise is
-  cleared as it settles, so a later subscriber reads current rows rather than a window that has
-  been drifting since boot. The result lands **in the lane and never backwards**: a snapshot that
-  resolved after a newer change was already fanned out is discarded and its caller is served from
-  the newer window, because rewinding hands that subscriber rows the fanout has moved past and a
-  cursor behind the change that would have corrected them.
-- The retained change window stores **pre-policy** patches; resume re-filters them per subscriber.
-- A resume is the one gate pass that runs **outside** the lane, and it reads the live window on
-  purpose: the window can only have moved forwards, and a row whose grant was revoked in the
-  meantime is one that pass must refuse rather than replay from its state at the cursor's lsn.
+## Numbers and ceilings
+
+- **Every numeric option is refused when it is not FINITE** — `@ultimat3/core`'s `finiteOption()`.
+  `bun run finite-bounds` reads zero for this package, but it cannot see an option with no `??`
+  default or where a screen runs, so audit by reading the option interfaces
+  (`SubscriptionBook`'s `maxPerTenant` and `openNatsClient`'s `maxReconnectAttempts` were the two it
+  missed). `SyncGrant.expiresAt` is app DATA, not an option, and is deliberately unscreened.
+- **A ceiling is refused where the object is BUILT**, never in a per-connection callback:
+  `socketCeilings()` (`sync-node-bounds.ts`) runs once in `createSyncNode`; `SyncSocket` keeps its own
+  screen because it is exported. `AcceptBudget`, `SyncSocket` and `SocketRegistry` throw
+  `X_INVARIANT` at construction on `NaN`/`±Infinity`.
+- **A ceiling per resource, and the wire's are not options** (table in `README.md`): the accept
+  budget bounds the rate, `maxConnections` the count, `socket.frameBudget` an open socket (checked
+  at the top of `routeFrame` **before `touch()`**), and `FRAME_LIMITS` in `decode` everything a
+  client sizes (narrowable, never widenable). `list()` takes a required `max`. `input` is walked
+  ITERATIVELY.
+- **Every `SyncSocket` ceiling is reachable from `createSyncNode`**, forwarded as
+  `...(x === undefined ? {} : { x })`.
+- **A `SubscriptionLimitError` names the knob**: every throw site passes `knob` explicitly;
+  `channel.test.ts` asserts the hub's two against the option names.
+- **Retained memory is bounded by BYTES**: `RingChangeBuffer` keeps the patch count as a replay
+  bound and byte budgets as the memory bound. `forget(qid)` runs when the last subscriber goes and
+  from `#dropIfUnheld` after a failed cold subscribe (only if the entry is still in the table, unheld
+  and has no read in flight).
+- **Every question a hot path asks is indexed**: `SubscriptionBook` keeps `#bySocket` and a
+  per-tenant count; `SocketRegistry` keeps `#byTopic`, changed only by `joinTopic`/`leaveTopic`.
+  `reauthorize` calls `book.retenant(socket)`.
+- **A cap is a RESERVATION taken before the first await**: `SubscriptionBook.reserve(socket, sid)`
+  decides the sid claim, `maxPerSocket` and `maxPerTenant` in one synchronous step;
+  `ChannelHub.subscribe` does the same for its caps. Released in a `finally`, idempotently.
+- **Readiness AND the connection cap are functions on `UpgradeDeps`**, re-asked after
+  `authenticate` (app code with an await) and right before `server.upgrade`. The recheck sheds with
+  the same 503 + `retry-after-ms` and spends no second `tryAccept()`.
+
+## Live queries
+
+- Policy is evaluated **once per subscriber**, never once per query (`live-query.test.ts`, and
+  `live-definition.test.ts` for a real `query({ live: true })`). Every policy call in `live-query.ts`
+  takes a `Subscriber`.
+- **A name nothing registered is `X_LIVE_QUERY_UNKNOWN`**, fix `x queries list --json`. The
+  registry is never enumerated over the wire.
+- **One build per `(query, input)`**: `target.live()` returns the descriptor and runs the read
+  (`LiveQuery.execute`), never memoised.
+- `liveQueryDefinition` caches per query id the compiled source, shape, matcher and shared window —
+  never a decision. The shared half is built with `enforce: false` on purpose; `authorize` runs per socket.
+- **The row policy sees the WHOLE row from the shared window.** A patch whose row the window does not
+  hold is WITHHELD (not denied, not a failure). A delta resume onto an unread entry fills it first
+  (`entry.lsn === ''`).
+- **A denial is a decision; everything else is a failure.** `visibleWithPolicy` matches
+  `QueryDeniedError` and rethrows the rest; `subscriber-gate.ts` and `reauthorize` ask
+  `isPolicyDenial(error)`. A failed snapshot raises out of `subscribe`; a failed delivery desyncs that
+  one subscriber; a failed `reauthorize` keeps the subscription. Failures count as `gateFailures` via
+  `onGateFailed`, never `onRowDenied`.
+- **One serial lane per query id (`WindowLock`) is the only thing that orders a fanout.** `deliver`
+  enters every lane before awaiting any; no fanout takes a second lane; a lane that fails desyncs its
+  own subscribers; lanes chain on a settled shadow of each task.
+- **Two reads of one entry are ordered by a READ GENERATION**, never by an lsn (`QueryEntry.lsn` may
+  be `''`): `entry.generation` / `entry.applied` is an identity check. The lsn guard stays for the
+  other question — a read that resolved behind a change the fanout already folded.
+- **The definition's read is once per entry**: a cold subscriber joins the in-flight read (a share,
+  cleared on settle), and the result lands in the lane and never backwards.
+- **The shared read carries a deadline and frees the SLOT**: `startRead` races it against
+  `DEFAULT_READ_DEADLINE_MS` (30 s; `new LiveQueryRegistry({ readDeadlineMs, schedule })`), rejects
+  joined callers with `X_TIMEOUT` (never an empty window), and puts `stale` back. No "off" spelling;
+  a non-positive or non-finite value takes the default (`query-window.test.ts`).
+- The retained change window stores **pre-policy** patches; resume re-filters them per subscriber,
+  outside the lane, against the live window.
+- **`desynced` is a mark with a reader**: the next delivery serves a fresh snapshot out of the shared
+  window and only then clears it. **`result.refill` is checked BEFORE the mark** — a guessed window
+  never clears a mark (`live-fanout.test.ts`).
+- **A change the window already holds is refused** (`change.lsn <= entry.lsn`, counted as
+  `staleChanges`). **A gap is detected**: the replicator stamps `producer` + `seq`; a skipped
+  sequence marks every window stale and every subscriber desynced; `refillWindowInLane` replaces it.
+- **A qid is `@ultimat3/query`'s `queryHash(name, input)`**; this package owns no hash
+  (`live-contract.test.ts`). The canonical form is core's and injective (`-0` is wire-reachable).
+- **A `sid` is CLIENT data: a subscription is keyed by `(socket, sid)`**
+  (`subscription-book.ts` is the only spelling). Reusing a sid the same socket holds is
+  `X_SUBSCRIPTION_ID_TAKEN`.
 - Truth is the server. A client is never the merge authority.
-- Presence lives in `transport.shared`, never in a node's heap — it must survive a node loss.
-- The `sync` node is `PresenceRegistry`'s only caller. Subscribing to a topic **is** joining its
-  presence set, repeating the frame is the heartbeat, and dropping the subscription or closing the
-  socket is the leave — presence has no frame of its own in either direction, because a second way
-  to say "I am here" is a client that can be subscribed and invisible at the same time.
-- Expiry is silent by design, so the node sweeps on an interval: without it a member whose node
-  died is never announced as gone, and the survivors render a cursor that stopped moving. It has to
-  be an interval — a sweep only reports members a previous sweep saw, which is how a member that
-  joined on another node becomes leavable here at all.
-- **The wire is the library's, the integration is ours, and `nats-client.ts` is the line between
-  them.** `nats` is imported by exactly one file — `nats-lib-client.ts` — and everything else in the
-  package, the transport and the JetStream bucket and the KV presence set and every test, is written
-  against that port. It is what lets the fake be an in-memory bus with *server* semantics instead of
-  431 lines of forged wire bytes, and what made deleting 1,019 LOC of framing, parser, PING/PONG and
-  session a swap rather than a rewrite ([`docs/idea/18-build-vs-wrap.md`](../../docs/idea/18-build-vs-wrap.md)).
-  The consequence that has to be held: **reconnect and re-subscription are the library's job now.** A
-  subscription outlives a dropped connection because the client re-establishes it underneath the
-  caller, so `NatsTransport` must never re-grow subscription bookkeeping of its own — a map of wanted
-  subjects, a dial promise, a loss counter, a rebind loop. Two things re-subscribing is a doubled
-  delivery on every reconnect and a subscription the caller's `unsubscribe` no longer reaches — the
-  hand-rolled client's lifecycle bugs were deleted with it rather than fixed for exactly that reason.
-  What stays above the port is what the library has no opinion on: our thundering-herd jitter, handed
-  over as its `reconnectDelayHandler` because spreading a restart herd is the framework's decision,
-  and the KV semantics presence needs (`nats-jetstream.ts`, `nats-kv.ts`) — a per-message TTL and a
-  batch direct get, which the library's own KV abstraction cannot express.
-- **Nothing leaves `NatsTransport` uncoded, and `#translating` is where that is enforced — added
-  2026-08.** `client.publish` and `client.subscribe` are the port's two SYNCHRONOUS calls, and the
-  library refuses locally on both: a bad subject, a payload over the server's `max_payload`, a
-  connection torn down between the `#ensure` and the call, a permissions violation on the subject.
-  A raw `NatsError` escaped `publish()` into `ChannelHub`'s bridge, `SocketRegistry` and the
-  replicator — no code, no `fix:`, nothing an operator can act on — while `InProcessTransport`
-  answered `X_TRANSPORT_UNAVAILABLE` for its own one refusal. `transport-parity.test.ts` asserts
-  both transports in one test, and a third case proves the wrap still DELIVERS: a translation that
-  swallowed a working publish would satisfy the two refusal cases and fan out nothing. An
-  `UltimateError` passes through untouched — the port raises its own for a closed client, and
-  re-wrapping buries the code a caller branches on. `nats-lib-client.ts`'s header claim that every
-  failure leaves *there* as an `UltimateError` is still false for those two calls; the translation
-  is deliberately in ONE place, and it is the transport, because `connect` is a public injection
-  seam and an app-supplied client throws whatever it likes.
-- **Reusing a client whatever its state is a DECISION, not the other half of that bug.** `#ensure`
-  hands back a client that is mid-reconnect on purpose: the library is re-establishing that same
-  connection and its subscriptions, and a second dial alongside it doubles every delivery. What a
-  caller gets from a client whose reconnect budget is spent is a publish that resolves into
-  nothing — reported through `onError` by `#watch`, and visible to `/readyz` through `connected`,
-  which is where a dead bus is meant to be caught.
-- One place reads `NATS_URL`, and it is `selectTransport` — a boot that resolved the bus itself
-  could reach a different one than the container it is standing in for. The KV bucket and the
-  presence TTL come back with the transport for the same reason: they are one decision.
+
+## The node, the bus and presence
+
 - `sync` is stateless: no sticky sessions, nothing on a socket survives a restart.
-- **A socket the node evicts ITSELF is released through `teardown`, never through
-  `sockets.remove`.** Bun's `close` callback runs `teardown`; a drain and the idle sweep have no
-  callback behind them — Bun's fires a tick later and `sockets.get` misses by then — so whatever
-  they do instead *is* the whole release. `drain()` inlined three of `teardown`'s five steps
-  (`close`, `sockets.remove`, `grants.delete`) and skipped the two the rest of the fleet can see:
-  `registry.unsubscribeSocket` and `presence.leave` per topic. What that left is a `QueryEntry`
-  whose `subscribers` map never empties — matcher, shared window and `WindowLock` pinned, and
-  `source.forget(qid)` never called — and, worse because it is cross-node, a presence member every
-  other node renders for a full TTL. During a **rolling restart** that is every room showing each
-  user twice for up to 30s, beside the same client's reconnection under a new socket id. One
-  `evict(socket, code, reason)`, and every path that ends a socket without a callback takes it.
-- **A `drain()` WAITS for the presence leaves it started; a `close` callback cannot** (2026-08-23).
-  `teardown` returns those promises as well as `detach`ing them: Bun's `close` callback is
-  synchronous, so there the detach is the whole of it — but a drain has no callback behind it and
-  is the one path that can wait. It did not: `release()`, `hub.close()` and the process's exit all
-  ran under N·M in-flight KV writes, so every other node rendered every drained member for a full
-  TTL — the same rolling-restart double vision `evict` exists to prevent, reached the long way
-  round. `evictInChunks` (`drain-evictions.ts`) evicts `DRAIN_EVICT_CHUNK` sockets, waits out what
-  they started, then takes the next: one synchronous loop over 50,000 sockets opens a quarter of a
-  million writes on one connection at the exact moment the fleet is already restarting.
-  `allSettled`, never `all` — a leave that fails is a member left to its TTL, which is what the
-  write meant when nobody waited for it at all, and it must not hold up the sockets behind it.
-- **The idle sweep exists, is armed by `start()`, and its budget is an APPLICATION one.**
-  `SocketRegistry.sweepIdle` had no caller for as long as it existed, so `touch()`, `idleFor` and
-  the 120s default decided nothing and `idleTimeoutMs` was unreachable from `createSyncNode`. The
-  only live guard was `websocket.idleTimeout: 120` handed to Bun — which Bun's own ping/pong
-  renews, so a client whose frame loop is wedged answers pings and keeps its `GrantBook` entry,
-  its `SubscriptionBook` entries and its `#byTopic` membership forever. It is now
-  `SocketRegistry.idle()`, a **query**: this table is three of the five things a socket holds, so
-  the object that can evict one is the node and not the registry. `start()` arms one `.unref()`ed
-  pass every `idleSweepPeriodMs(idleTimeoutMs)` — a quarter of the budget, floored at a second,
-  derived rather than configured because a second knob is a second number that can disagree with
-  the one it is a fraction of — and `release()` clears it beside the presence sweep. **It measures
-  on `Clock.monotonic()`**, the clock `AcceptBudget` already uses: the sweep compares a DURATION,
-  and a duration read off `now().getTime()` is decided by whatever NTP last wrote — a step forward
-  evicts every socket that is talking, a step backward makes `idleFor` negative and spares every
-  socket that is dead, and the sweep had only just gained its first caller when both became
-  reachable. The field is named `lastSeenMonotonicMs` so nobody hands it to `new Date()`;
-  `openedAt` is the wall-clock one and stays that way, because a human reads it.
-- **`drain()` and `stop()` both release what `start()` acquired, and releasing twice is a no-op.**
-  A `drain()` is terminal on its own — it closes the hub and evicts every socket — and nothing
-  obliges a `stop()` to follow it, so leaving the change subscription and the presence sweep to
-  `stop()` alone is a drained node still pulling every change off the bus into a fanout with no
-  sockets, and still sweeping a room it left, through a hub it already closed. One `release()`,
-  called by both. `drain()` calls it after the sockets are gone and before `hub.close()`: a client
-  is entitled to its patches for the whole grace window, and to get them through a hub that is
-  still open.
-- Exactly one `replicator` per DB, enforced by a session-level advisory lock.
-- **A start and an acquisition are MEMOISED, because `running` and `#connection` are both written
-  after an await (`As of 2026-09-06`).** `start()` read `if (running) return true` and then awaited
-  `lock.tryAcquire()`, so two overlapping calls both passed the guard, both were told they held the
-  lock — a holder's `tryAcquire()` answers `true` — and both ran `feed.start()`: one replication
-  slot, two pumps, every change published twice under two `seq` generations of one producer id,
-  which every sync node's `SeqGapDetector` reads as a gap and repairs by re-snapshotting the fleet.
-  `PgAdvisoryLock.tryAcquire` had the identical hole one layer down and a worse ending: two
-  concurrent calls opened two SESSIONS, the second assignment to `#connection` orphaned the first,
-  and `release()` closes one — so the key stayed held until the process died and no standby could
-  ever take the slot. Both are now the shape `packages/core/src/lifecycle.ts`'s `drain()` uses: a
-  synchronous guard-and-register, the promise published before anything is awaited, and the memo
-  cleared however it settles (a `false` is "somebody else holds it right now", which the takeover
-  loop asks again a backoff later). `stop()` and `release()` await the in-flight one rather than
-  reading their own flag — that flag is false for the whole of a start, so an unguarded teardown
-  returns "nothing to do" and leaves behind exactly what it was called to release.
-  **A start that FAILS hands the lock back, and `running` is set after the feed is pumping**
-  (`As of 2026-09-06`). It was set before `await feed.start()`, so a feed that rejected — a slot
-  already `active`, a preflight refusal — left this node holding the advisory lock and claiming to
-  run with nothing pumping: the takeover loop's next `start()` was answered `true` by the
-  `if (running)` guard without re-entering `begin`, and every standby stayed a standby of a slot
-  whose holder was not replicating. The release is best-effort, because the feed's failure is the
-  one the operator acts on and a session-scoped lock a dead connection cannot release is released
-  by Postgres when that session ends.
-- **The replication pump has one way out, and it closes what it held.** Both exits — a decode error
-  and `nextCopyData()` returning `undefined`, which is the walsender ending the copy — run `#die`:
-  record `stats().failure`, stop the confirm timer, close the connection and null it. Each one left
-  behind is a dead replicator claiming to be a live one. A `null` failure answers `/readyz` ready
-  for a loop reading no WAL; a live `#running` makes the next `start()` a silent no-op; a retained
-  timer keeps telling the walsender a dead stream is keeping up; a retained `#connection` is a
-  socket the next `start()` overwrites rather than closes, holding the slot `active`. A `start()`
-  that goes live clears the previous failure, and `stop()` awaits the pump even when the connection
-  is already gone — `#die` nulls it *before* closing it, so returning early reports a released slot
-  to the supervisor that is about to start the next process.
-- **`#pump` *is* the terminal cleanup, so a restart awaits it before it dials.** `#drain` awaits
-  `#die` and `#die` awaits `connection.close()`, but `#die` clears `#running` and nulls
-  `#connection` before that close settles: a `start()` checking `#running` alone dialled into a
-  slot the dead walsender still owned and replaced `#pump` with its own, so the next `stop()`
-  awaited only the new pump. `start()` takes the previous pump and awaits it first; its failure
-  path calls `stop().catch(() => undefined)` because the boot diagnosis is the one an operator acts
-  on and a teardown that also failed must not replace it.
-- **`stop()` releases everything before it reports anything.** A `#confirm` or an `endCopy` that
-  threw skipped the close and the pump await, leaking the socket and telling a supervisor the
-  teardown was over before it had begun. Every step runs whatever the step before it did, and the
-  first failure is rethrown only once the connection is closed and the pump has ended.
-- **`REPLICA IDENTITY FULL` is asked about at preflight, warned about, and counted — never thrown**
-  (added 2026-08-19). `pg-replication.ts`'s `#deliver` hands a `delete` its `message.before`,
-  and under any identity but FULL that tuple is the KEY COLUMNS ALONE — which `toRow` accepts,
-  because it only requires a text `id`. So `bridgeChange` decided "did this row leave the result
-  set" from a one-column row, `visible` read `undefined` for every column the identity did not
-  carry, and nothing anywhere recorded that it had happened: no emit, no check, and
-  `X_LIVE_REPLICA_IDENTITY` existed in neither the source nor the manifest. The check is a fourth
-  `connection.query` in `preflight`, and it **must** stay ahead of
-  `pg_create_logical_replication_slot` — changing the identity after a slot exists does not reach
-  what that slot decodes. It WARNS (`logger.warn(code, { cause, fix, tables })`, the message being
-  the code alone) because a throw would stop every app on the default identity from booting, which
-  is a worse outcome than the partial rows. `ReplicationStreamStats.partialBefore` is the runtime
-  half, read off `PgRelation.replicaIdentity` and never off the tuple: a DEFAULT-identity table
-  whose non-key columns happen to be NULL sends the bytes a FULL one does, so counting absent keys
-  would undercount exactly the rows a policy is most likely to misjudge. A hard refusal in the
-  `x verify` step is the follow-up and lives in `@ultimat3/cli`.
-- **The replication session pins its own output formats, `As of 2026-08-23`.** Postgres sends every
-  WAL value as TEXT and `pg-values.ts` reads a `timestamptz` by matching postgres' ISO spelling,
-  keeping the raw text when it does not match — deliberately, because a wrong instant is worse than
-  a string. That makes the SERVER's `DateStyle` load-bearing: `SQL`, `German` or `Postgres` sends
-  every timestamp down the fallback, the shared window holds `Date`s while the patch holds text,
-  `compareValues` falls to string comparison, and one edit to one column jumps its row to the top
-  of every `orderBy('createdAt','desc')` feed for every subscriber — the exact defect the decode
-  exists to close, re-opened by a GUC. `pg-connection.ts` therefore sends
-  `options: '-c datestyle=ISO -c intervalstyle=postgres -c extra_float_digits=3'` in the startup
-  packet, byte for byte what postgres' own logical-replication client sends
-  (`libpqwalreceiver.c`) — which is why a walsender accepts it. On **every** session this class
-  opens, not only the replicating one: one session shape is one thing to reason about, and the
-  advisory-lock connection is the same class. A server that refuses one answers `ErrorResponse` at
-  startup, so the replicator fails to boot with the server's own words rather than mis-sorting a
-  feed behind a warning nobody reads. `pg-connection.test.ts` pins the packet.
-- A change lsn is `<16 hex commit position><8 hex row position in that transaction>`. Never order by
-  either half alone: the commit lsn repeats within a transaction, and per-record WAL positions are
-  not monotonic across transactions. Never make it depend on wall time, the entity list or a process
-  counter — a replay must produce byte-identical lsns or at-least-once turns into duplicate delivery.
-- Slot, publication and entity names are interpolated into a replication command, so they are
-  checked against `[a-z_][a-z0-9_]*` first. That regex is a security boundary, not a style rule.
-- Same rule on the bus, for the half that is still ours: a bucket name is interpolated into a
-  JetStream stream name and its API subjects, so it is checked first (`assertBucket` in
-  `nats-jetstream.ts`, `X_TRANSPORT_PROTOCOL`). Subject validation went with the hand-rolled client —
-  the library refuses a malformed subject itself, and a second spelling of that rule here is a second
-  place it can drift. A presence key or member id is user data, so it is base64url-encoded
-  (`encodeToken`) rather than validated — no name is refused for its spelling.
-- **One record store per PAGE, and `record-store.ts` is the only place a record lives (21.0.0).**
-  Keyed `type:key` — the entity's NAME and the key the SERVER computed with the entity's own
-  projection (`live-record-type.ts`, via `recordProjectionForTable`): an envelope arrives keyed, a
-  live snapshot carries `keys` and a patch `key` whenever the record key is not the row's `id`, and a
-  non-live `useQuery` window is the key order of the answer's `records[type]`. The browser never
-  derives a key; an answer with no envelope holds its own rows, not records. Two layers: SYNCED is server truth, the OVERLAY is every pending optimistic write,
-  REPLAYED over synced truth whenever it moves — never a before-image restored over newer truth,
-  which is what the old journal did. A live window, a `useQuery` list and a `useRecord` are all
-  projections over it; none holds a row of its own. It is core's `RecordSink`, installed on
-  `pageClient().store`, so an HTTP answer adopts into it without this package in the call.
-- **Page state lives on `globalThis[Symbol.for('ultimate.realtime')]` (`page-store.ts`), never in
-  module scope.** Every island is its own bundle with its own copy of this package, so a module
-  singleton is one store and one socket PER ISLAND — the bug the plan-101 major ended.
-  `page-client.test.ts` builds this package twice with `Bun.build` and proves one store and one
-  `new WebSocket` between the copies. Never `instanceof` across copies: the class is whichever
-  bundle's copy made it.
-- **The signal factory is per BUNDLE, the store per PAGE.** Each island carries its own solid-js, so
-  a signal must come from the bundle whose effects read it: `installRealtime({ signal, sync })`
-  (`reactivity.ts`, module scope on purpose) is what the island bootstrap calls. No install and a
-  DOM is `X_REALTIME_UNINSTALLED`; no install and no DOM is a server render — every hook answers
-  `pending`/online and creates NO page state (on a server that would be one store for every
-  request). `useRecord` and `useQuery` return `AsyncState` accessors the caller releases.
-- **One socket per page, built by the first live hook (`page-socket.ts`), and `browser-socket.ts`
-  holds the framework's only `new WebSocket`.** A form-only or `useRecord`-only island never
-  imports that module, so it ships none of the lifecycle — measured, a `useRecord`-only chunk
-  carries no `WebSocket`. `import()` would buy nothing: islands build with `splitting: false`, and
-  Bun inlines a dynamic import there. `hasPageSocket()` lives in `page-store.ts` so asking costs no
-  socket bytes either. A principal change (`rescope`) clears the store and redials.
-- **The socket is READ-ONLY (protocol 3, 21.0.0).** `mutate`/`rebase` frames and
-  `createSyncNode({ onMutate })` are deleted; an old client's `mutate` decodes to
-  `X_PROTOCOL_VERSION`. A write is `useMutation`: the twin into the overlay, then `POST
-  actionPath(name)` through core's `clientTransport` with an idempotency key; the answer's records
-  are adopted INSIDE the transport call, before `settle` drops the overlay, so a convergent twin
-  never flickers. A refusal drops the overlay. An `ack` is now only ever a refusal: `ref` is a
-  subscription's sid (that window renders `failed`) or the socket (reported through `onError`).
-- **`LiveClient` holds no signal.** It is shared by every bundle on the page, and a signal belongs to
-  one bundle — so status is plain reads plus `onStatus`, and a live window is `LiveHandle` reads
-  plus `onChange`. Hooks wrap both in the calling bundle's own signal.
-- **An answer that did not carry a row the write touched does not end its overlay.** `useMutation`
-  settles against the answer's own records (`onEnvelope`); a touched row it did not carry (an
-  action that returns a view, not the entity) keeps its overlay until the server's next row for it
-  — a frame, another answer — bounded by `DEFAULT_AWAIT_SERVER_MS` (10 s), after which the synced
-  layer stands. Without it, a like showed `+1`, fell back, then rose again on the frame.
-- **A `records` frame names the write that produced it, and the writing page settles on it
-  (21.0.0, additive to protocol 3).** The node fans a commit out before it answers, so the write's
-  own frame routinely beats its HTTP answer, and merged under the still-pending twin it painted
-  the write twice: measured in `examples/dummy`'s `offline-like` e2e, `2,2 → 3,3 → 2,2` on the
-  outbox replay. No client-side rule can tell its own echo from somebody else's change without the
-  frame saying which write it is, and deferring frames for touched rows was rejected: with no
-  version it reorders an older frame over a newer answer. `ChannelRecordsFrame.write` is
-  `writeDigest(idempotencyKey)` (`@ultimat3/core`), never the key, because a frame goes to every
-  member. `RecordStore.push` digests every overlay key it takes (`record-names.ts`), before the
-  request leaves. `client-channels.ts` calls `settleWrite` inside the frame's own batch, so the
-  merge and the settle are one notification. A digest this page does not hold changes nothing.
-  Server side, the name comes off `ChangeEvent.write`: `channel-logs.ts` stamps it on the ring
-  entry, so a `since` replay names it too. `pg-replication.ts` reads it off the transaction's
-  opening `pg_logical_emit_message` (prefix `WRITE_ORIGIN_WAL_PREFIX`, START_REPLICATION asks
-  `messages 'true'`), and `@ultimat3/testing`'s in-process replicator reads it off the row
-  observer. A frame `write` that is not a digest is a protocol error (`wire-channel.ts`); on the
-  bus it is dropped, never trusted (`parseEnvelope`). `write-echo.test.ts` covers the useMutation and
-  outbox paths, another writer, an unknown digest, a custom policy, and a partial echo.
-- **An overlay a server answer partly confirmed stops painting on those rows.** `settle` with rows
-  missing used to keep the WHOLE twin, so the rows the answer or echo did carry showed the write
-  twice until the rest arrived. The carried rows now go into `OverlayEntry.confirmed`, and
-  `replayOverlays` leaves them to synced truth. Only the rows still waiting show the twin.
-- `local(tx, input)` is pure and CONVERGENT: no I/O, no `Date.now()`, no `Math.random()`, and
-  applying it over its own result changes nothing. The overlay replays it on every server update.
-- Anything a component reads is a **getter or an accessor**, never a value snapshotted at hook time:
-  a plain field cannot re-render. `MutatorLike.local` is declared with method syntax so a twin
-  typed over its own `tx` assigns with no cast.
-- `useQuery`'s input is read once, at call time. There is no reactive runtime here to re-run it.
-- Every subscription handle client code gets back — `LiveHandle` (`subscribeLive`), the
-  `useQuery`/`useRecord` accessors, `Unsubscribe` (`client.subscribe(topic, …)`) — is `Disposable`,
-  and `[Symbol.dispose]` is the exact same function reference as the release, never a second
-  teardown path. Pinned in `type-pins.ts`.
-- Type claims about the hook go in `type-pins.ts`, never in a `.test.ts` — `tsconfig.json` excludes
-  test files, so `tsc -b` never reads one and an assertion written there can never fail.
-- **`backoffDelay` is `@ultimat3/core`'s, and `attempt + 1` is the whole of the seam** (`As of
-  2026-08-23`). A client counts its first reconnect as attempt **0** and core counts the first wait
-  as attempt **1**, so dropping the shift doubles every reconnect delay in the framework —
-  silently, and only under the load this file exists to survive.
-  `thundering-herd-core-parity.test.ts` pins it with the numbers (`[500, 1_000, 2_000, 4_000]`) and
-  with 17,280 comparisons across every jitter mode, base, cap, factor, attempt and roll.
-  `JitterMode` is core's type re-exported and `Rng` is core's `Random` — never a second spelling of
-  either. `drainPlan()` and `AcceptBudget` are NOT backoff and keep their own arithmetic: a slot in
-  a spread and a token bucket's refusal delay are different questions. `bun run flight-copies` is the
-  guard: a second curve-and-jitter function anywhere in `packages/*/src` is a build error, matched
-  on the literal shape rather than the name.
-- **The SHARED window read carries a deadline, and it frees the SLOT — it cannot cancel the read**
-  (`As of 2026-08-23`). One `definition.snapshot` that never settled pinned `entry.reading` for the
-  life of the process, and every later cold subscriber joined a promise nothing would ever resolve:
-  one wedged read took every future subscriber of that query id with it. `startRead` now races the
-  read against `entry.schedule(…, entry.readDeadlineMs)`, default `DEFAULT_READ_DEADLINE_MS` (30s),
-  injectable per entry and through `new LiveQueryRegistry({ readDeadlineMs, schedule })`. Three
-  rules, each with its own test in `query-window.test.ts` and each proven by mutation:
-  - **A race, not just an eviction.** Freeing the slot alone leaves every caller ALREADY joined
-    awaiting a promise nothing settles; they are told instead, with `X_TIMEOUT`.
-  - **`X_TIMEOUT`, not a silent empty window.** A superseded read is discarded silently because a
-    strictly better window already exists to serve from; a timed-out read has none, and serving
-    `rows: []` as the whole result set is the exact fault `live-query.ts`'s `#read` refuses. The
-    rejecting-read path (`readSnapshot`'s catch) is the shape this matches, not the superseded one.
-  - **`stale` is put back**, for `readSnapshot`'s reason: a read that did not answer must not leave
-    the window looking authoritative.
-  There is **no "off" spelling** — a shared read with no deadline is the defect itself — and a `0`,
-  negative, `NaN` or `Infinity` value falls back to the default rather than to "now".
-- The client owns its own reconnect: a closed socket arms **one** timer through the injected
-  `Scheduler`, and that timer calls `connect()`. `reconnectAt` is the render half and never the
-  mechanism — publishing it without arming anything is exactly the bug that shipped. Rules that
-  hold the arming together: `onClose` nulls `#socket` (a retained dead socket makes `#send` a
-  silent no-op), it only schedules when nothing is armed (a `reconnect` frame arms the node's
-  spread slot *before* closing, and a local backoff would overwrite it), and `close()` cancels —
-  a client whose owner is gone must stop dialling, while `connect()` starts it over. A close speaks
-  only for **its own** socket: `onClose` returns before touching any state when `#socket` is no
-  longer the socket that closed, because a replaced socket closing late must not mark the live
-  connection offline or arm a backoff behind it — and `close()` therefore reports its subscriptions
-  offline itself. A dial that throws inside the timer arms the next attempt and is **reported
-  through `onError`** (default `console.error`; never `logger`, whose writer is `process.stderr`
-  and this is browser code): a socket constructor may refuse, one refusal ending the chain is the
-  same outage as never arming, and nothing awaits a timer — a throw out of one is an uncaught
-  exception that can kill the process that was going to retry. Only the timer owns the chain and
-  only the timer reports — a `connect()` the app called itself throws to the app and arms nothing.
-- **A `sid` is CLIENT data, so a subscription is keyed by `(socket, sid)` — never by `sid` alone.**
-  `LiveQueryRegistry.unsubscribe(socketId, sid)` and `.subscription(socketId, sid)` both take the
-  owner. Keyed by the sid alone, socket B reusing socket A's sid overwrote A's slot: A's
-  subscription stayed in its query entry's `subscribers` map with nothing able to reach it, so
-  `unsubscribeSocket(A)` freed nothing, `subscribers.size` never hit zero, and the entry's matcher
-  and shared window were pinned for the process's life while every change fanned out to a dead
-  socket. A `{op:'drop', sid}` frame from B ended A's stream with no error on either side.
-  `sync-node` passes `socket.id` on the drop path for that reason. Reusing a sid the SAME socket
-  already holds is `X_SUBSCRIPTION_ID_TAKEN` — refused rather than replaced, because replacing is
-  the strand. `subscription-book.ts` owns that identity and is the only place it is spelled: the
-  query entry's own `subscribers` map takes the same composite key, so one `unsubscribe` reaches
-  both by one identity.
-- **`connect()` closes the socket it is replacing, and a frame speaks only for its own socket.**
-  A remount calling `connect()` on a live client left the previous socket open: its `onMessage`
-  kept folding patches into the live registrations, and the node held two sockets for one client —
-  double presence membership, double fanout — until the tab closed. `#socket` is nulled before the
-  close so the corpse's `onClose` takes its early return, and `onMessage` carries the same identity
-  guard `onClose` already had.
-- **The grant is recorded BEFORE `server.upgrade`, and released on the path that never opens.**
-  Bun runs `websocket.open` SYNCHRONOUSLY inside `server.upgrade` and does not return until it has
-  (bun 1.4.0), and `open` is where `sync-node` reads the `GrantBook` for the socket's actor.
-  Recorded on the line after the upgrade — which it was — every authenticated socket on a real node
-  carried `actor: null`: `ChannelHub.#authorize` denied every topic with `X_TOPIC_FORBIDDEN`,
-  `authorize`/`visible` decided about nobody, `maxPerTenant` never applied and `hello.actorId` was
-  null. It did not self-repair, because `GrantBook.expired()` skips a grant with no `expiresAt` —
-  the shape `authenticate: async () => ({ actor })` produces. `onUngranted` is what makes the
-  correct order safe (only a `close` callback deletes a grant, and an upgrade that never took gets
-  no callback); it is REQUIRED on `UpgradeDeps`, so a second host of `handleUpgrade` cannot forget
-  it. The harness is half the rule: `sync-node-auth.test.ts`'s `upgradeTarget()` returned `true`
-  without ever calling `open`, so the bug was invisible to every test here — it now opens the socket
-  inside `upgrade()`, the way Bun does.
-- **Every `socket.send` on the node reads its answer, and what a `false` costs is decided per
-  frame.** A subscribe reply is REPAIRABLE and was the silent one: `registry.subscribe` has already
-  seated the subscription and cleared its desync mark, so a dropped snapshot left the server
-  believing a client holding no rows was in sync, and every later change reached it as a patch
-  folded onto nothing — on a healthy socket, forever. It is marked desynced, exactly as
-  `live-fanout` marks a lost patch. A `rebase`/`ack` has nothing to mark (the node keeps no
-  per-mutation state) and a client only returns an `inflight` mutation to its queue when the
-  connection dies, so an undeliverable settlement **closes the socket**: the reconnect requeues and
-  replays it under the same idempotency key, and acking a rebase that never left is the
-  rebase-before-ack order defeated one frame later. A presence roster has no repair at all — the
-  membership is already on the shared set and the client's next heartbeat re-rosters — so it is
-  logged (`sync.presence_roster_dropped`). The drain's `reconnect` frame is the socket's slot in the
-  spread and nothing re-sends it, so `drain()` returns `DrainedSocket[]` with `notified` per socket
-  and logs `sync.drain_frames_dropped`.
-- **A socket's actor comes from `createSyncNode({ authenticate })` and from nowhere else.** The node
-  imports no authenticator — the app supplies one — and it runs
-  on the upgrade *before* `server.upgrade`, so a refused credential never costs a websocket.
-  `null` is a **decision** (401, `X_SOCKET_UNAUTHENTICATED`, a client fault that pages nobody); a
-  throw is a **failure** (503, `X_SOCKET_AUTH_UNAVAILABLE`, reported) — the same rule the row gate
-  follows, one layer out. Absent, every socket is anonymous and `start()` warns: that node is
-  single-tenant, and `hub.guard('org.*.feed', ({ actor }) => actor?.orgId === …)` denies everyone.
-  The actor is written in exactly one place, the `GrantBook`; `WsData` deliberately carries none,
-  because two spellings of one identity disagree the moment a re-auth renews one of them.
-- **A grant expires; a socket does not.** `authenticate` answers a `SyncGrant`, not an `Actor`: a
-  15-minute token on a socket that stays up for hours was authorized once and served forever, and
-  an active client never idles out either — every inbound frame `touch()`es it. The node re-decides
-  an expired grant on an interval and then calls both halves that already existed and had no
-  caller: `hub.onActorChange` (topics) and `registry.reauthorize` (subscriptions). `refresh` is the
-  app's closure, so the framework retains no credential of its own — re-reading the upgrade
-  `Request` would mean holding one per socket. No `refresh` = close with `1008` and let the client
-  re-dial. A `refresh` that **raises** keeps the socket and retries: a token service timing out is
-  not a revocation.
-- **`desynced` is a mark with a reader.** It is written when a patch is dropped by backpressure,
-  when a gate fails, when a window loses its tail and when a re-auth survives; the *next* delivery
-  serves that subscriber a fresh snapshot out of the shared window (no DB read) and only then
-  clears it. A snapshot the socket refuses leaves the mark, which is the state it is in. Four
-  writers and no reader was a subscription that stayed permanently and silently stale on a healthy
-  socket, with the server knowing and the client not.
-- **`result.refill` is checked BEFORE the mark, because a repair out of a guessed window clears it.**
-  The word "fresh" above is load-bearing: when the matcher lost the window's tail, `entry.rows` is a
-  guess, and the same fanout that refuses to send a *patch* derived from it was resnapshotting every
-  already-desynced subscriber out of it — and clearing the one mark that would have made the next
-  change re-read. That subscriber is then recorded as repaired against rows nothing trusts and gets
-  a patch, not the snapshot it is still owed, from the refilled window. A lost tail degrades every
-  subscriber the same way, whatever each was holding, and they are all repaired on the next change
-  after `refillWindowInLane` has replaced the window. `live-fanout.test.ts` pins both halves.
-- **A change the window already holds is refused on the way in.** The replicator guarded duplicates
-  and out-of-order on the *publish* side; `entry.lsn = change.lsn` was unconditional on the
-  *consume* side, so a redelivery rewound every subscriber's cursor. `change.lsn <= entry.lsn` is
-  dropped and counted as `staleChanges`.
-- **A gap in the change stream is detected, not assumed away.** Fanout is core NATS — at most once
-  — and an lsn cannot reveal a gap, because a WAL position is a byte offset and every legitimate
-  next change is already an arbitrary jump. The replicator stamps `producer` + `seq`; a skipped
-  sequence marks every window `stale` and every subscriber desynced, and the next change to each
-  query re-reads. Both fields are optional on the bus: a publisher that does not sequence detects
-  nothing rather than crying gap, and a *new* producer restarts the count rather than reading as
-  one. A stale window is replaced in the lane (`refillWindowInLane`) — `fillWindow` takes the
-  entry's own lane and a lane is not reentrant.
-- **A hub that closed opens nothing, and `#open` is the only thing that can enforce it.** `close()`
-  walks `#bridges` and then clears it, which reaches every bridge that is open and none that is
-  still opening: a reservation an in-flight `subscribe` has taken is `sub === null`, so
-  `unsubscribeWhenOpen` does nothing to it and `clear()` drops the entry. The transport then hands a
-  live subscription to a `Bridge` nothing can name — `#release` looks the topic up, misses and
-  returns — and its handler keeps calling `deliver` for the life of the process. The same orphan the
-  `Bridge` comment describes, one state earlier. So `close()` sets `#closed` **before** the walk and
-  `#open` closes its own subscription when it lands after one, dropping the entry with it so a
-  second post-close subscribe opens and closes its own rather than double-unsubscribing this handle.
-  It then **raises** `X_TRANSPORT_UNAVAILABLE` rather than returning: returning let `subscribe` fall
-  through to `joinTopic`, so the socket became a member of a topic nothing on this node is bridged
-  to — silent for the life of the connection, no error on either side, and no reason for the client
-  to redial. Reachable between `hub.close()` inside `node.drain()` and the last in-flight subscribe.
-  `#release` takes the bridge the caller reserved for the same reason: after `close()` cleared the
-  table, that topic name may hold a bridge a LATER subscribe opened, and releasing by name alone
-  decrements somebody else's refcount.
-- Deny by default on topics. No guard = `X_TOPIC_FORBIDDEN`.
-- **A guard that FAILS is not a guard that denied — the hub's copy of the rule the row gate already
-  follows.** On `onActorChange` (the re-auth pass) only a denial unsubscribes; anything else keeps
-  the topic, increments `guardFailures` and logs `channel.guard_failed`. A guard is app code and may
-  reach a database, so `catch { unsubscribe }` reported a store that timed out as a revoked grant —
-  every topic on every re-authenticated socket on the node, silently, with the client never told to
-  resubscribe. The initial `subscribe` is deliberately NOT split: there is no subscription to keep,
-  so a raising guard refuses that subscribe and the client hears about it.
-- **One conflict vocabulary, and it is not declared here.** A mutator's `conflict` is
-  `ConflictPolicy` from `@ultimat3/core`, over ROWS, and `RecordStore.settle` hands it to core's
-  `resolveConflict` with the overlay's row and the adopted server row — in that order. Until 21.0.0
-  there were two spellings bridged by a filter in `hooks.ts` that dropped one and handed the other
-  a `{ local, base, server }` bag, so no merge an app declared ever decided a row. A server delete,
-  or a row the client never held, is not a conflict: the server's answer stands, no merge runs.
-- **Inbound frames run in a lane, and the lane is NEVER the socket.** `sync-node.message` dispatches
-  every frame as `void (async () => routeFrame(…))()`, so nothing upstream orders them. A global
-  per-socket lane would put every frame behind the slowest one, and the slowest one is a subscribe's
-  snapshot read — a DB round trip every reconnecting client pays once per live query, which is the
-  restart storm this framework is measured on. `mutate` is one lane per socket, `subscribe` is
-  `sub:<sid>` or `topic:<name>`, everything else is unlaned (`frame-lanes.ts`). A lane exists only
-  while work is queued on it: keyed by a client-chosen sid, a lane that outlived its work is an
-  unbounded map one socket grows at will.
-- **A cap is a RESERVATION taken before the first await, never a check.** A lane makes concurrent
-  frames sequential and N sequential subscribes still pass a check-then-act cap N times — and the
-  per-tenant cap spans sockets, where no lane can see it at all. `SubscriptionBook.reserve(socket,
-  sid)` decides the sid claim, `maxPerSocket` and `maxPerTenant` in one synchronous step;
-  `ChannelHub.subscribe` does the same for `maxTopicsPerSocket`, `maxTopicsPerNode` and the node's
-  bridge slot, before the guard is awaited. The tenant is captured, not re-derived — a re-auth may
-  `retenant` the socket while the read is in flight, and the release has to give the slot back to
-  the tenant that took it. Released in a `finally`, and releasing twice is a no-op.
-- **Bun's native pub/sub is deleted, not wired.** Nothing here publishes to a native topic and
-  nothing will: a native publish cannot be refused per socket, cannot report the frame it dropped
-  and cannot mark a subscriber desynced. `SocketRegistry.deliver` is the one fanout path.
-  `WsLike.subscribe`/`unsubscribe` stay declared and unused — a tracked app implements the
-  interface structurally, so removing the members is that app's typecheck failure — and the
-  declaration says so, because a member that looks live is one someone will call.
-- **A dropped `records` frame is counted AND repaired (plan 101, slice 10).** Every declared
-  channel carries a per-node `seq` and `epoch`; a frame backpressure refuses marks that socket
-  gapped on that channel, and the node sends `replay-gap` the moment the socket drains (the
-  websocket `drain` handler → `gapRepairs.repairAll`), counted as `channel_replay_gaps_total` beside
-  `channel_frames_dropped_total`. The client re-runs the channel's catch-up query, holds every frame
-  that arrives meanwhile, and applies those over the read. The server owns the gap verdict: a
-  numeric hole on its own is never one (a row this socket may not see is skipped for it). A resume
-  sends `since` = the highest seq with no hole below it; duplicates are dropped, a replay that
-  fills a hole is applied (`client-channels.ts`).
-- **A channel is a DECLARATION, never a topic string.** `channel(name, { params, policy, catchUp,
-  records?, events? })` spells the topic; a subscribe names the declaration and its params, and a
-  name no declaration carries is `X_TOPIC_FORBIDDEN`. Presence rides the channel's `events` frames
-  (`readPresence`) — only a channel declared `events: true` has a room. `useChannel` / `usePresence`
-  hold one membership per topic per page, however many components ask.
-- **One socket per ORIGIN and principal, in a SharedWorker (plan 101, slice 11).** `socket-engine.ts`
-  holds the real socket and talks to every tab only over a `MessagePort`; to each tab's `LiveClient`
-  its port IS a socket (`socket-host.ts`'s virtual socket). Channel wants are reference-counted
-  across ports and every server frame is ROUTED to the ports that want it; live-query sids carry the
-  port. A tab's own beat is the engine's ping: a port silent for `REAP_AFTER_BEATS` is a closed tab
-  and is reaped. When the real socket drops, every tab's virtual socket closes and each tab
-  resubscribes from its OWN cursors — the engine keeps nothing that could go stale. No
-  `SharedWorker`, a constructor that throws, or no built worker: the in-page host runs the same
-  engine over a `MessageChannel`, one socket per tab. The worker is named by principal, so two
-  principals never share a socket.
-- **A qid is `@ultimat3/query`'s `queryHash(name, input)`, and this package derives none of its
-  own — `As of 2026-08`.** `qidOf` was the same two lines over a local copy of the canonical form
-  (`stableDigest(canonicalJson(input))`), and `canonicalJson`/`stableDigest` were this package's
-  third copy of what `@ultimat3/action` and `@ultimat3/query` also each held. They had already
-  diverged: `{ a: undefined, b: 1 }` gave `feed:eb8ed3ccb5023093` from `queryHash` and
-  `feed:c0bf82ad036cb0a5` from `qidOf`, because query's walk drops an `undefined`-valued key and
-  this one rendered `"a":null`. The two are COMPARED in one flow — `@ultimat3/query`'s `planResume`
-  decides refetch-vs-resume by comparing a cursor's `queryHash` against the query's, while
-  `liveQueryDefinition` keys the shared window by the qid — so keeping both correct was never the
-  option; the first time either moved, every resume decision and every window lookup were keyed
-  differently. `realtime -> query` is the one declared sideways edge and this package already
-  imports it. **`fnv1a` is gone too, `As of 2026-08-24`**: it stayed for one job — the cursor's
-  result-set digest — and `LiveCursor.digest` was deleted for having no reader, so this package now
-  owns no hash at all. A 32-bit hash nothing calls is one the next caller reaches for as a sharing
-  key, which is the single thing `json.test.ts` used to pin it against.
-  `live-contract.test.ts` is the pin — it reads `registry.subscriberCount(queryHash(name, input))`
-  through a real subscribe, so a local derivation fails it. Cost of the move: none observable on the
-  server. Every qid a node computes comes from a DECODED frame, and `JSON.parse` produces no
-  `undefined`, no `Date`, no `Map` and no `Set` — the four values the two forms disagree about — so
-  no live subscription re-keyed and nothing re-snapshotted.
-- **The canonical form is injective over the values it accepts, and `JSON.stringify` is not** —
-  the reason that survives the move, now `@ultimat3/core`'s to enforce. `JSON.stringify` answers
-  `"null"` for `NaN` and `±Infinity` and `"0"` for `-0`, so four distinct inputs hashed to one qid
-  — and a qid *hit* hands the joiner the first subscriber's compiled source, matcher and seated
-  window. Bare `NaN` / `Infinity` / `-Infinity` / `-0` tokens are emitted instead; they are not
-  valid JSON, which is correct, because that output is hashed and never parsed. Exposure is
-  narrower than it looks and the tests say so rather than overclaiming: `NaN` and `±Infinity` have
-  no JSON spelling and so cannot arrive on a `subscribe` frame — they reach the hash only from a
-  caller building `input` in JS. **`-0` is wire-reachable**: `JSON.parse('{"a":-0}')` answers `-0`.
-- **Refusing new sockets and draining the ones you have are two shutdown phases.** `stopAccepting()`
-  is the `accept` phase: `ready = false`, `/readyz` 503, a late upgrade shed with `retry-after-ms`,
-  and every socket untouched — a draining node still owes its clients their patches, and `stop()` is
-  what releases the change subscription carrying them. `drain()` + `stop()` are the `close` phase.
-  Registered with no phase, both landed in `close` and the node upgraded new websockets until the
-  very end. `listenSyncNode` unregisters both on `stop()`.
-- **Readiness AND the connection cap are asked twice, because `authenticate` is app code with an
-  await in it.** A request that passed the checks at the top of `handleUpgrade` can be parked in a
-  token service when SIGTERM lands, and the `accept` phase is over by the time it reaches
-  `server.upgrade` — one more socket on a node the load balancer has already stopped routing to, so
-  nothing takes it over. `ready` and the socket count are therefore **functions** on `UpgradeDeps`,
-  not values read once.
-  **`socketCount()` was the half that was read once and never re-asked** (2026-08-23), which is the
-  same staleness with a worse blast radius: a restart storm dials every client of a dead node at
-  this one at once and each parks in the token service having passed the cap while the node still
-  held nothing, so `maxConnections: 2` with ten parked upgrades took **ten** sockets — reproduced,
-  `upgraded 10, shed 0`. Sound because there is no await between the recheck and `server.upgrade`,
-  and the count moves INSIDE it: Bun runs `websocket.open` synchronously there, which is where
-  `sockets.add` runs. The recheck sheds with the same 503 + `retry-after-ms` and takes no second
-  `tryAccept()`: that budget was spent.
-- **A client `send` that returned is not an acknowledgement.** A browser `WebSocket.send` on a
-  CLOSING socket discards the frame and returns normally, so a drained mutation is `inflight` until
-  the server settles it or `requeueInflight` returns it. Only `pending` is sendable, `drain()` is one
-  chained pass at a time (two overlapping passes put one key on the wire twice, and a later pass can
-  overtake the one ahead of it), and backpressure over `MAX_BUFFERED_BYTES` declines rather than
-  fails — the mutation stays pending and the pass stops instead of reordering the ones behind it.
-- **The lane orders passes; it does not order a socket death, so the queue carries an epoch.**
-  `requeueInflight` is not a pass and cannot reach into one parked at `await send(...)`: it hands
-  back what was on the dead socket, the parked pass resumes and marks everything *behind* that
-  mutation `inflight` for a connection that is gone. `#sendable` excludes `inflight`, so the next
-  drain skips them, no ack ever arrives and the writes are lost — invariant 3 inverted. `#epoch` is
-  bumped before the requeue scan and read at the top of every `#pass` iteration; a pass whose epoch
-  went stale returns and leaves the rest `pending` for the connection that arms the next one.
-- **`#persist` hands the store a SNAPSHOT, never the live entries.** `QueueStore.save` is a durable
-  write (OPFS, IndexedDB) and may await before it reads. Given the array itself, a store that
-  resolves after the next pass has moved on persists a status that was never true when it was
-  called — and `inflight` is the one a reload cannot recover from.
-- **A reconnect replays registrations AND topics, and every socket handler carries the identity
-  guard.** A reconnect is one `hello` plus one frame per thing this client holds: a `subscribe` per
-  registration, carrying that registration's cursor, and a `subscribe` per topic. Topic membership
-  is state on the node's socket, so a channel is silent from the first reconnect while its handler
-  is still installed — and its presence membership is swept — unless every one is re-announced.
-  `onOpen` needed the `#socket !== socket` guard `onMessage` and `onClose` already had: a replaced
-  socket opening late marked the connection up and replayed every subscription onto the current one.
-- **`hello` carries NO cursors, and `HelloFrame.resume` is deleted (2026-08).** It was filled by
-  every client on open and read by nobody — the node replied `resume: []` and decided resume per
-  subscription from the `subscribe` frame — so every reconnect shipped each cursor twice, up to 512
-  ids each, in the restart storm this package is measured on. Wiring it was the wrong half of the
-  choice: a cursor's `qid` is `` `${name}:${fingerprint(input)}` ``, so a node reading a resume list
-  recovers the query **name** — it is the plaintext prefix — but never the `input`, which is the half
-  every decision needs. Without it `definition.authorize({ actor, input })` cannot run and no entry
-  can be built; the qid names a window but not a decision, and the retained window holds pre-policy
-  patches, so answering from it at `hello` time means answering before the per-subscriber
-  authorization pass, for a subscription that does not exist yet. It could only ever restate,
-  unauthorized, what `subscribe` decides with the input in hand — and it could not even save the
-  bytes, because the cursor still has to ride its `subscribe`. Two places deciding one thing is what axiom 1 refuses. **`PROTOCOL_VERSION` did NOT
-  move**, same rule as `snapshot.entity`: `decode` is a whitelist, so a new node drops an old
-  client's `resume` and an old node reads a new client's omission as the empty list it always got.
-  The one deploy of skew costs nothing in either direction.
-- **The client beats, because only the client can end a half-open socket.** `heartbeatMs` (default
-  `DEFAULT_HEARTBEAT_MS`, 15s; `0` disables) sends a `hello` — byte-identical to the opening one,
-  since the frame has no resume list to leave out — plus one subscribe frame per topic, which is the
-  node's presence heartbeat. It is **not** how a deploy is noticed: `socket.skewed` compares the
-  build the client claims (the `hello`'s `buildId`, which `sawHello` records on every one — the
-  latest is the record — or `?build=` on the dial) against this node's; a client says the same
-  build on every beat and the node's never moves while the socket is open, so every `hello` on one
-  socket answers the same forever and `update-available` reaches a client on the
-  socket it opens against the *new* node. The hello IS read — until 2026-09-07 only the dial was,
-  and a dial without `?build=` was recorded as this node's own id, so a client naming its build only
-  in the frame was current forever. Two silent windows and the client closes with `4000` and
-  arms the reconnect. It is one
-  self-re-arming tick on the injected `Scheduler`, not an interval: a client is either beating on a
-  live socket or backing off toward a new one, never both. The 15s is the client's OWN number:
-  `realtime.heartbeatMs` was a `RealtimeConfig` key read by nothing and it is **deleted**
-  (2026-08-19). The server half of the beat stays derived — `PresenceRegistry.heartbeatMs` is
-  `max(1000, floor(ttlMs / 3))`, the same rule `idleSweepPeriodMs` follows, because a second knob
-  is a second number that can disagree with the one it is a fraction of.
-- **Every question a hot path asks is indexed, never scanned.** `SubscriptionBook` keeps
-  `#bySocket` and a per-tenant count beside `#bySid`, and `SocketRegistry` keeps `#byTopic` beside
-  the socket table. Both replaced a walk of the whole node that ran once per socket or once per
-  frame: `ofSocket` filtered a copy of every subscription (100,000 entries measured at **17.7s** of
-  blocking work per teardown or re-auth sweep — a deploy or a batch of grants expiring together is
-  the whole trigger), and the per-tenant cap walked the same map on **every subscribe frame**
-  (7.96 ms each at that size). A new index goes where the deaths are seen: topic membership is the
-  registry's because `remove` is the one path a close, a drain and the idle sweep all take, and
-  `joinTopic`/`leaveTopic` are the only way to change it — two call sites for one membership is how
-  an index goes wrong. When an actor changes, `reauthorize` calls `book.retenant(socket)`: an index
-  nobody updates is a count that drifts for the rest of the process.
-- **A ceiling per resource, and the wire's are not options.** `README.md` has the table. The rule
-  behind it: the accept budget bounds the accept *rate*, so the *count* needs its own
-  (`maxConnections`, shed as the same 503 + `retry-after-ms`); a socket that is open needs a frame
-  budget (`socket.frameBudget`, checked at the top of `routeFrame` **before `touch()`** — a frame
-  this node refuses must not renew the idle window); and anything a client sizes is bounded in
-  `decode` by `FRAME_LIMITS`, which a caller may narrow but never widen. `list()` takes a required
-  `max` so a new array field on a new frame cannot ship without someone choosing its size.
-  `input` is walked ITERATIVELY: the thing being refused is a stack overflow in `canonicalJson`, so
-  a recursive check would be the same crash one frame earlier.
-- **Every ceiling on a socket `sync` builds is reachable from `createSyncNode`.** The node
-  constructs every `SyncSocket` it holds, so a `SyncSocketOptions` the node does not forward is a
-  number an operator can only change by abandoning `createSyncNode` — which is what
-  `maxBufferedBytes` and `maxDroppedFrames` were until 2026-08. Forwarded the same way
-  `maxFramesPerSecond`/`frameBurst` already are (`...(x === undefined ? {} : { x })`, so an unset
-  option keeps `SyncSocket`'s own default rather than overwriting it with `undefined`).
-- **A `SubscriptionLimitError` names the knob, never the default.** `knob` defaults to
-  `maxPerSocket`/`maxPerTenant`, which are `LiveQueryRegistry`'s — so the channel hub's per-socket
-  *topic* cap, thrown without one, told an operator to move a number in a different constructor
-  that would not have helped. Every throw site passes `knob` explicitly (`maxTopicsPerSocket`,
-  `maxTopicsPerNode`, `maxEntries`, `maxPerSocket`, `maxPerTenant`); `channel.test.ts` asserts the
-  two hub ones against the option names, because a fix line naming the wrong setting is worse than
-  no fix line — it is an instruction that runs and changes nothing.
-- **Retained memory is bounded by BYTES.** `RingChangeBuffer` keeps the patch-count cap as a
-  *replay* bound (what a delta resume costs to fold) and adds the byte budgets as the memory one —
-  `packages/cache/src/lru.ts:1-2` states why: 4,096 queries x 1,024 patches is 4.19M retained rows
-  and no number of bytes at all. `forget(qid)` is called by `LiveQueryRegistry.unsubscribe` when the
-  last subscriber of a query id goes; it had no caller, so the ring outlived the entry. It is
-  also called by `#dropIfUnheld` on the subscribe path, `As of 2026-09-06`: an entry is created
-  BEFORE the snapshot read that fills it, so a cold subscribe the database refused left an entry
-  with no subscriber and no removal path — `unsubscribe` can only reach one through a subscription
-  that was never attached. `maxEntries` cold failures were therefore enough to answer
-  `X_SUBSCRIPTION_LIMIT` to every later subscriber for the life of the process, after the database
-  had recovered, because `qid` derives from client-chosen input and distinct inputs mint distinct
-  orphans. Dropped only when the entry is still the one in the table, holds no subscriber and has
-  no read in flight — a read published on it belongs to a concurrent subscriber that has not
-  attached yet, and dropping it there would hand that subscriber a window no change reaches.
-- **A channel patch id carries the NODE that minted it** (`As of 2026-09-06`). `ChannelHub`'s
-  `#sequence` counts within one process, and the patch id was that counter alone — so two `sync`
-  replicas publishing to one topic minted the same id for the same subscriber, and a channel has
-  no cursor and no re-snapshot, so nothing downstream can repair a collision. `nodeId` defaults to
-  a per-hub `uuid()` and is declarable (the pod name) when an operator should be able to say which
-  node published a frame. The frame's `lsn` is deliberately left as the per-process counter:
-  nothing reads a channel frame's lsn as an order across nodes, and `client-frames.ts` advances a
-  cursor only for a registered live query, never for a topic.
-- **An error never renders a value that carries a credential.** `parsePgUrl` names `DATABASE_URL`
-  rather than echoing the URL it refused — an error reaches a log, `--json`, an agent transcript
-  and a ticket, and the password is in the string. Same rule as `packages/mail/src/driver-smtp.ts`.
-- **A full presence frame is capped and says so; the set behind it is never capped.** `roster()` is
-  what a frame carries (`maxMembers`, 256, plus `total`); `list()` stays whole because the sweep
-  differences it, and a short list would report every member past the cap as having left. `total`
-  is set on `sync` only — a `join`/`leave`/`update` frame is a delta, and a count beside one reads
-  as truncation.
-- **One node per topic sweeps.** Every node sweeping every room it has seen is one full-set read
-  multiplied by the fleet, and the same `leave` frame published N times. The election needs no
-  compare-and-set the shared store does not have: the lease key is a *keyed set*, so each node's
-  claim is its own member and the leader is the lowest id every claimant can see. Eventually
-  consistent on purpose — the worst case is a duplicate `leave` for someone already gone.
-- **Money is THREE physical columns on the wire too, and a live row must equal a repository row.**
-  `entityRow` folds `<p>_minor`/`<p>_currency`/`<p>_scale` into one property. It matched two, so a
-  scaled amount arrived at every subscriber unscaled *and* carrying a stray physical `priceScale`
-  beside `price` — one row, two shapes, no error anywhere. NULL and absent both mean **no `scale`
-  key**, never `0` (that is whole units, a 100x reinterpretation of an ordinary price), which is
-  exactly what `@ultimat3/entity`'s `moneyOf` does. That equality is the pin:
-  `pg-entity-row-parity.test.ts` reads one physical row through both surfaces — this package's fold
-  and a real `postgresRepo` — and asserts one object, each side absolutely as well as against the
-  other, because equality alone is satisfied by both failing open together. It is the one test here
-  that imports `@ultimat3/entity` (tier 2, a legal downward edge, test-only: `*.test.ts` never
-  ships), and it has to, or the thing being compared against is a copy of the reader instead of the
-  reader. The `0…15` scale bound stays `@ultimat3/schema`'s — enforced by the column CHECK and by
-  `parseScale`, never restated here.
-- Never a bare `Error`. Never `any`. Never `Date.now()` — take a `Clock` (`clock.now()` is a `Date`;
-  use `monotonic()` for durations).
-- **A test fixture standing in for a FOREIGN error extends `Error` on purpose, and that is not the
-  bare-`Error` rule being broken.** `PoolTimeout`, `Denied`, `MutationFailed` and `ThirdPartySdkError`
-  (nine sites across `realtime`, `db` and `ai`) simulate a driver, a policy library or an app's
-  `onMutate` — values this package did not construct and must handle anyway. `isPolicyDenial`,
-  `stringField` and `renderThrowable` all exist *because* such values arrive; rebuilt as
-  `UltimateError`s the fixture would prove the framework handles its own errors, which is the
-  "equality satisfied by both sides failing open together" failure the row-parity test names. The
-  rule governs what this package **throws**, never what a test hands it.
+- **`selectTransport(env, realtime)` is the one place `realtime.transport` + env become a bus**; the
+  KV bucket and presence TTL come back with it. The config decides, the env supplies (22.0.0):
+  `'nats'` dials the variable `urlEnv` names and refuses (`X_CONFIG_INVALID`) when unset; `'memory'`
+  refuses a set `NATS_URL` (or the named variable).
+- **`nats` is imported only by `nats-lib-client.ts`**; reconnect and re-subscription are the
+  library's job — `NatsTransport` must never grow subscription bookkeeping. What stays above the
+  port: the herd jitter (`reconnectDelayHandler`) and the KV semantics presence needs
+  (`nats-jetstream.ts`, `nats-kv.ts`). `nats-fake.ts` is a bus with server semantics.
+- **Nothing leaves `NatsTransport` uncoded** — `#translating` wraps the port's synchronous
+  `publish`/`subscribe` refusals as `X_TRANSPORT_UNAVAILABLE`; an `UltimateError` passes through.
+  `transport-parity.test.ts` asserts both transports, and that the wrap still delivers. `#ensure`
+  reuses a mid-reconnect client on purpose.
+- **A bucket name is checked before interpolation** (`assertBucket`, `X_TRANSPORT_PROTOCOL`); a
+  presence key or member id is base64url-encoded (`encodeToken`), never validated.
+- Presence lives in `transport.shared`. The sync node is `PresenceRegistry`'s only caller:
+  subscribing to a topic IS joining, repeating the frame is the heartbeat, dropping or closing is the
+  leave. Expiry is silent, so the node sweeps on an interval, and **one node per topic sweeps** (lowest
+  id in the lease key's keyed set). `roster()` caps a frame at `maxMembers` (256) with `total`;
+  `list()` is never capped. `total` is set on `sync` frames only.
+- **A socket the node evicts ITSELF goes through `evict(socket, code, reason)`** (the full
+  `teardown`), never `sockets.remove`. **`drain()` waits for the presence leaves it started** —
+  `evictInChunks` (`drain-evictions.ts`, `DRAIN_EVICT_CHUNK`, `allSettled`).
+- **The idle sweep is armed by `start()`** (a quarter of `idleTimeoutMs`, floored at 1 s, `.unref()`)
+  and measures on `Clock.monotonic()` (`lastSeenMonotonicMs`). `SocketRegistry.idle()` is a query;
+  the node evicts.
+- **`drain()` and `stop()` both call one idempotent `release()`**; `drain()` releases after the
+  sockets are gone and before `hub.close()`.
+- **Refusing new sockets and draining are two phases**: `stopAccepting()` is `accept`, `drain()` +
+  `stop()` are `close`. `listenSyncNode` unregisters both on `stop()`.
+- **A hub that closed opens nothing**: `close()` sets `#closed` before the walk; `#open` closes a
+  late subscription and RAISES `X_TRANSPORT_UNAVAILABLE`. `#release` takes the reserved bridge, never
+  a name.
+- **A socket's actor comes from `createSyncNode({ authenticate })` only**, run before
+  `server.upgrade`. `null` = 401 `X_SOCKET_UNAUTHENTICATED`; a throw = 503
+  `X_SOCKET_AUTH_UNAVAILABLE`. Absent = anonymous, and `start()` warns. The actor lives only in the
+  `GrantBook`.
+- **The grant is recorded BEFORE `server.upgrade`** (Bun runs `open` synchronously inside it) and
+  released on every path that never opens — `onUngranted` is REQUIRED on `UpgradeDeps`, and a
+  `server.upgrade` that THROWS releases too. `sync-node-auth.test.ts`'s harness opens inside
+  `upgrade()`, as Bun does.
+- **A grant expires; a socket does not**: expired grants are re-decided on an interval through
+  `hub.onActorChange` and `registry.reauthorize`. No `refresh` = close with `1008`; a `refresh` that
+  raises keeps the socket and retries.
+- **Every `socket.send` reads its answer**: a dropped subscribe reply marks the subscription desynced;
+  a dropped presence roster is logged (`sync.presence_roster_dropped`); `drain()` returns
+  `DrainedSocket[]` with `notified` and logs `sync.drain_frames_dropped`.
+- **Inbound frames run in a lane, never the socket**: `subscribe` is `sub:<sid>` or `topic:<name>`,
+  everything else unlaned (`frame-lanes.ts`); a lane exists only while work is queued.
+- **Bun's native pub/sub is deleted**; `SocketRegistry.deliver` is the one fanout path.
+  `WsLike.subscribe`/`unsubscribe` stay declared for structural implementers.
+- **A dropped `records` frame is counted AND repaired**: per-node `seq`/`epoch` per channel; a refused
+  frame marks the socket gapped, and the websocket `drain` handler sends `replay-gap`
+  (`channel_replay_gaps_total` beside `channel_frames_dropped_total`). The client re-runs the
+  channel's `catchUp`, holding frames meanwhile.
+- **A channel is a DECLARATION**: `channel(name, { params, policy, catchUp, records?, events? })`.
+  Deny by default, and `policy` is REQUIRED (`X_CHANNEL_DECLARATION_INVALID`; a public channel says
+  `policy: allow('public')`). Presence rides `events: true` channels.
+- **A guard that FAILS keeps the topic on re-auth** (`guardFailures`, `channel.guard_failed`); the
+  initial subscribe still refuses.
+- **A channel patch id carries the node** (`nodeId`, default a per-hub `uuid()`).
+- **An error never renders a credential**: `parsePgUrl` names `DATABASE_URL`, never the URL.
 
-## Browser bytes, per hook (`As of 2026-09-22`)
+## Replication
 
-`bun build --target=browser --minify --metafile`, one entry importing one hook from the barrel.
-Re-measure before quoting.
+- Exactly one `replicator` per DB, by a session-level advisory lock. **`start()` and
+  `PgAdvisoryLock.tryAcquire` are MEMOISED** (guard and promise published synchronously, cleared on
+  settle); `stop()`/`release()` await the in-flight one. A start that FAILS hands the lock back, and
+  `running` is set only once the feed pumps.
+- **The pump has one way out, `#die`**, which records the failure, stops the confirm timer, closes and
+  nulls the connection. `start()` awaits the previous pump before it dials. **`stop()` releases
+  everything before it reports anything.**
+- **`REPLICA IDENTITY FULL` is checked at preflight, warned, and counted — never thrown** — ahead of
+  `pg_create_logical_replication_slot`. `ReplicationStreamStats.partialBefore` reads
+  `PgRelation.replicaIdentity`, never the tuple.
+- **Every session pins `datestyle=ISO`, `intervalstyle=postgres`, `extra_float_digits=3`** in the
+  startup packet (`pg-connection.ts`, pinned by `pg-connection.test.ts`), byte for byte what
+  Postgres' own walreceiver sends.
+- A change lsn is `<16 hex commit position><8 hex row position>`; never order by either half alone,
+  never depend on wall time or a process counter.
+- Slot, publication and entity names match `[a-z_][a-z0-9_]*` before interpolation — a security
+  boundary. A SQLSTATE is data: `pg-wire.ts`'s `FIXES` is read with `Object.hasOwn`.
+- **A live row equals a repository row**: `pg-entity-row.ts` decodes through `@ultimat3/entity`'s
+  own `decodeRow` / `entityForTable`, so money (three physical columns), scale and every other kind
+  fold exactly as `postgresRepo` folds them.
+- A write's name rides the WAL: `pg-replication.ts` reads it off the transaction's opening
+  `pg_logical_emit_message` (prefix `WRITE_ORIGIN_WAL_PREFIX`; `START_REPLICATION` asks
+  `messages 'true'`).
 
-| Hook | Before the page boot moved | After the page boot moved | After the light core entry (current) |
-|---|---|---|---|
-| `useRecord` | 34,838 | 19,404 | **13,226** |
-| `useMutation` | 37,073 | 35,480 | **29,907** |
-| `useQuery` | 64,183 | 62,596 | **56,647** |
-| `useChannel` | 62,155 | 60,566 | **54,621** |
-| `@ultimat3/realtime/boot` (once per page, cached immutable) | — | 34,884 | not re-measured |
+## The page
 
-The last column (`As of 2026-09-22`, `bun build --target=browser --minify`, one entry per hook)
-is after every browser-reachable file here moved its core imports to **`@ultimat3/core/page`** and
-`@ultimat3/query/client` stopped reaching `@ultimat3/query`'s `errors.ts`: no hook loads a titles
-table any more — not core's (`core-error-codes.ts`), not schema's, not query's. A browser file here
-imports core from `@ultimat3/core/page`; `packages/core/src/page-bundle.test.ts` and
-`packages/query/src/client-bundle.test.ts` are the guards.
+- **One record store per PAGE (`record-store.ts`)**, keyed `type:key` with the SERVER's key
+  (`live-record-type.ts`); the browser never derives one. SYNCED truth plus an OVERLAY of pending
+  optimistic writes, REPLAYED over synced truth whenever it moves. Installed as core's `RecordSink` on
+  `pageClient().store`.
+- **Page state lives on `globalThis[Symbol.for('ultimate.realtime')]` (`page-store.ts`)**, never module
+  scope — every island is its own bundle (`page-client.test.ts` builds the package twice). Never
+  `instanceof` across copies.
+- **The signal factory is per BUNDLE**: `installRealtime({ signal, sync })` (`reactivity.ts`). No
+  install and a DOM is `X_REALTIME_UNINSTALLED`; no DOM is a server render — hooks answer
+  `pending`/online and create no page state. `useRecord` and `useQuery` return `AsyncState` accessors.
+- **One socket per page, built by the first live hook (`page-socket.ts`)**; `browser-socket.ts`
+  holds the framework's only `new WebSocket`. `hasPageSocket()` lives in `page-store.ts`. A principal
+  change (`rescope`) clears the store and redials.
+- **One socket per ORIGIN and principal, in a SharedWorker**: `socket-engine.ts` holds it and talks
+  to tabs over `MessagePort`s (`socket-host.ts`'s virtual socket); wants are ref-counted and frames
+  routed; a port silent for `REAP_AFTER_BEATS` is reaped; each tab resubscribes from its OWN cursors.
+  No `SharedWorker` → the in-page host runs the same engine over a `MessageChannel`.
+- **The socket is READ-ONLY (protocol 3)**. A write is `useMutation`: the twin into the overlay, then
+  `POST actionPath(name)` through core's `clientTransport` with an idempotency key; the answer's
+  records are adopted inside the transport call, before `settle`. An `ack` is only a refusal.
+- **`LiveClient` holds no signal** — plain reads plus `onStatus` / `onChange`; hooks wrap them.
+- **An answer that did not carry a touched row keeps that row's overlay** until the server's next row
+  for it, bounded by `DEFAULT_AWAIT_SERVER_MS` (10 s).
+- **A `records` frame names the write that produced it** (`ChannelRecordsFrame.write =
+  writeDigest(idempotencyKey)`, never the key). `RecordStore.push` digests every overlay key before
+  the request leaves (`record-names.ts`); `client-channels.ts` settles inside the frame's own batch.
+  A non-digest `write` is a protocol error (`wire-channel.ts`) and dropped on the bus. Rows an answer
+  or echo partly confirmed go to `OverlayEntry.confirmed` and stop painting. `write-echo.test.ts`.
+- `local(tx, input)` is pure and CONVERGENT: no I/O, no `Date.now()`, no `Math.random()`.
+- **One conflict vocabulary**: `ConflictPolicy` from `@ultimat3/core`, resolved by core's
+  `resolveConflict(overlayRow, serverRow)` in `RecordStore.settle`. A server delete is not a conflict.
+- Anything a component reads is a getter or accessor. `MutatorLike.local` uses method syntax.
+  `useQuery`'s input is read once, at call time.
+- Every subscription handle (`LiveHandle`, the hook accessors, `Unsubscribe`) is `Disposable`, and
+  `[Symbol.dispose]` is the same function as the release. Type claims go in `type-pins.ts`.
+- **The outbox**: only `pending` is sendable; `drain()` is one chained pass at a time; backpressure
+  over `MAX_BUFFERED_BYTES` declines rather than fails. `#epoch` is bumped before the requeue scan and
+  read every pass iteration. `#persist` hands the store SNAPSHOTS by KEY (`QueueChange`); one tab
+  drains at a time (`navigator.locks`, `page-outbox.ts`'s `exclusive`) and resets a stored `inflight`
+  to `pending` under the lock.
 
-`examples/dummy`'s islands, built by `x build`'s own `buildIslands`: `likes-badge` 48,598 → 32,882,
-`feed` 95,983 → 94,174, `like` 84,351 → 82,503; `settings` and `contact-sales` reach no realtime
-and are unchanged.
+## The client connection
 
-- **The disk boot is ONE page script, never island code.** `boot.ts` (`./boot`) restores the
-  principal's persisted records and opens the outbox; the CLI builds it like the sync worker
-  (`/_x/page-boot/<hash>.js`, immutable) and renders one `<script defer>` on a document that carries
-  the scope tag AND emitted an island reaching realtime. The boot first wipes every stored scope
-  but the current principal's (rows and outbox) — a sign-out by full navigation never calls
-  `rescope()` — then restores; an unscoped page wipes nothing. Two principals in two tabs: the newer
-  boot wipes the other's disk, which keeps its records in memory and re-persists on its next write.
-  `page-store.ts` imports none of it: `booted` reads the boot's promise off
-  `globalThis[Symbol.for('ultimate.page-boot')]` per access, so an island that ran first still waits.
-  A lazy `import()` could not have done this: islands build with `splitting: false`, and Bun inlines
-  a dynamic import, so bytes leave a bundle only by leaving its import graph.
-- Everything else on these paths buys function: the in-page `socket-engine` is the no-SharedWorker
-  fallback, `client-channels` rides the ONE page client every bundle shares, the decoders are one
-  copy. Core's error-registry chain (~8.4 kB with the schema and query titles) WAS core's to cut,
-  and is cut: see the last column above.
-- **An island never carries the outbox.** `boot.ts` is the one module that builds it (IndexedDB,
-  the queue, the drain listeners); `useMutation` and the page socket read it off the page through
-  `outbox-slot.ts`, after `page.booted`. No boot ⇒ no outbox ⇒ a write the network refused is
-  rejected, not queued in memory. Measured the same day, same method, before → after:
-  `useMutation` 30,071 → 21,725, `useQuery` 56,824 → 47,696, `useChannel` 54,798 → 45,585;
-  `examples/dummy` via `buildIslands`: `feed` 101,261 → 92,973, `like` 86,325 → 78,050.
-- **A browser path never loads realtime's code table.** The refusals a browser can reach live in
-  `page-errors.ts`; `errors.ts` re-exports them and keeps the table and its one
-  `registerErrorCodes()`. Measured the same day, same method: `useRecord` 21,629 → 18,600,
-  `useMutation` 38,225 → 35,194, `useQuery` 65,964 → 62,935 (`useChannel` 60,903 after).
-  `page-errors-bundle.test.ts` fails if `errors.ts` comes back into `useRecord` or `useMutation`.
-  In a browser that loaded no table a code titles itself from its name; `code`, `cause` and `fix`
-  are unchanged. The ~7.6 kB of core's `UltimateError` chain that was left under `useRecord` is now
-  the class and the lookup alone (~2.4 kB): core's titles table is an anchor only its barrel loads.
+- **The client owns its reconnect**: a closed socket arms ONE timer through the injected `Scheduler`
+  that calls `connect()`. `onClose` nulls `#socket`, schedules only when nothing is armed (a
+  `reconnect` frame's slot wins), and speaks only for its own socket; `close()` cancels. A dial that
+  throws in the timer is reported through `onError` (default `console.error`) and arms the next.
+- **`connect()` closes the socket it replaces**; `onOpen`, `onMessage` and `onClose` all carry the
+  identity guard.
+- **A reconnect replays registrations AND topics** — one `hello`, one `subscribe` per registration
+  (with its cursor) and per topic.
+- **`hello` carries NO cursors** (`HelloFrame.resume` is deleted): a cursor's qid names a window but
+  not the input a decision needs.
+- **The client beats** (`heartbeatMs`, default `DEFAULT_HEARTBEAT_MS` 15 s, `0` disables): a `hello`
+  plus one subscribe per topic (the presence heartbeat). Two silent windows close with `4000`. The
+  node's `socket.skewed` compares the build the `hello` claims against its own. The server beat is
+  derived: `PresenceRegistry.heartbeatMs = max(1000, floor(ttlMs / 3))`.
+- **There is ONE `backoffDelay`, `@ultimat3/core`'s, counted from 1.** `policyDelay(policy, attempt,
+  rng)` (internal, not on the barrel) maps a `BackoffPolicy` onto it; a 0-based counter adds 1 at the
+  call site (`client.ts`, `socket-engine.ts`, `replicator.ts`'s 0-based `retryDelayMs`).
+  `client-reconnect.test.ts`, `socket-engine-reconnect.test.ts`, `nats-transport.test.ts` pin the
+  first wait at the base. `drainPlan()` and `AcceptBudget` keep their own arithmetic.
+  `bun run flight-copies` refuses a second curve.
 
-## Map
+## Rules for code here
+
+- Never a bare `Error`. Never `any`. Never `Date.now()` — take a `Clock` (`monotonic()` for durations).
+- **A test fixture standing in for a FOREIGN error extends `Error` on purpose** (`PoolTimeout`,
+  `Denied`, `ThirdPartySdkError`): it simulates a value this package did not construct. The rule
+  governs what this package **throws**, never what a test hands it.
+
+## Browser bytes, per hook (`As of 2026-09-22`)
+
+`bun build --target=browser --minify`, one entry importing one hook from the barrel. Re-measure
+before quoting.
 
-| File | Owns |
+| Hook | Current |
 |---|---|
-| `index.ts` | the `.` barrel: the client half, and the only thing a browser island may import |
-| `server.ts` | the `./server` barrel: the bus, the WAL path, the node. Disjoint from `index.ts` by test |
-| `sync-protocol.ts` | the wire: 10 frame kinds, `encode`/`decode`, `PROTOCOL_VERSION` |
-| `channel.ts` / `presence.ts` / `socket.ts` | tier 1 |
-| `live-query.ts` / `live-definition.ts` / `changefeed.ts` / `changefeed-env.ts` / `replicator.ts` / `pg-advisory-lock.ts` / `fanout.ts` / `transport-env.ts` / `matcher-bridge.ts` | tier 2 |
-| `pg-bytes.ts` / `pg-wire.ts` / `pg-auth.ts` / `pg-connection.ts` / `pg-socket.ts` | the Postgres v3 client: bytes, frames, SASL, session, socket |
-| `pgoutput.ts` / `pg-entity-row.ts` / `pg-replication.ts` | WAL decode → `ChangeEvent`, and the lsn that orders it |
-| `pg-preflight.ts` | the four questions asked before `START_REPLICATION` — `wal_level`, the publication, every entity's replica identity, the slot — plus `assertIdentifier`, the charset all four interpolate through |
-| `nats-client.ts` | the bus port: publish/subscribe/request/requestMany/close/version/connected, and `parseNatsUrl` — the library takes `host:port` plus credentials and never reads a URL's userinfo |
-| `nats-lib-client.ts` | the `nats` adapter — **the only file in the repo that imports `nats`** |
-| `nats-jetstream.ts` / `nats-kv.ts` / `nats-transport.ts` | the JetStream KV bucket, presence over it, and the production `Transport` — all three written against the port |
-| `nats-fake.ts` | an in-memory bus implementing the port — server semantics, not wire bytes; the only way to prove multi-node fanout under a sealed network |
-| `cursor.ts` / `change-buffer.ts` / `thundering-herd.ts` | reconnect — the highest-risk area. `thundering-herd.ts`'s backoff is core's, shifted 0-based to 1-based; the drain plan and the accept budget are its own |
-| `page-errors.ts` | the refusals a browser can reach — the store, the hooks, the wire check, the local store — without the code table |
-| `record-store.ts` | the page's one record store: the optimistic overlay over synced truth, batched notification, and `settle` under the conflict policy — from an answer, or from the write's own `records` echo (`settleWrite`) |
-| `record-names.ts` | which pending overlay a frame's `write` digest names: every pushed key digested before its request leaves |
-| `record-key.ts` / `record-synced.ts` / `record-await.ts` | a record's `type:key` name and `carriedBy`; the synced layer (merge, holds, provisional disk rows); the overlays waiting on server truth and what a write heard while in flight |
-| `record-tx.ts` | the overlay replay and the `tx` a mutator's `local` half writes through |
-| `page-store.ts` | the page state on `globalThis` — the store, the sync target, the write counts — and `hasPageSocket` |
-| `outbox-slot.ts` | the outbox as an island reaches it: the page slot, its handle type and `OutboxEntry` — read, never built |
-| `boot.ts` | `./boot`: the page's ONE boot script — the disk restore and the outbox open, never in an island |
-| `page-socket.ts` / `browser-socket.ts` | the page's one socket, built by the first live hook, and the framework's one `new WebSocket` |
-| `reactivity.ts` | `installRealtime`: this bundle's signal factory, and the server-render test |
-| `use-record.ts` / `use-query.ts` / `use-mutation.ts` / `use-connection.ts` / `use-channel.ts` | the hooks — the only surface an island calls |
-| `client-channels.ts` | the client's declared channels: one membership per topic, the seq/epoch cursor, the catch-up read and the frames held during it |
-| `socket-engine.ts` / `socket-host.ts` / `sync-worker.ts` | the one socket per origin: the port-driven engine (dial, beat, redial, reap), the tab's host choice and virtual socket, the SharedWorker entry (`./sync-worker`) |
-| `socket-port.ts` / `socket-routes.ts` | the engine ⇄ tab port messages; the multiplexing — one membership per topic, every frame routed only to the ports that want it |
-| `sync-meta.ts` | the sync target and worker URL read off the document's `<meta>` tags (core's names) |
-| `live-record-type.ts` | the live path's record type and record key per table, from the entity's own projection |
-| `live-rows.ts` | one live subscription's window over the store — its record type, its order, its retain/release, and `Registration` itself |
-| `offline-queue.ts` | the durable outbox queue; `page-outbox.ts` opens one per principal and replays it in order over HTTP |
-| `client.ts` / `sync-node.ts` | the two halves — connection lifecycle and subscriptions; the socket carries no writes |
-| `sync-auth.ts` | what a socket's identity IS (`SyncGrant`), the book that holds one per socket, and the pass that re-decides an expired one |
-| `sync-frames.ts` | what a RECEIVED frame does to server state — the node's inbound surface, and the mirror of `client-frames.ts` |
-| `sync-upgrade.ts` | the node's HTTP surface: `/healthz`, `/readyz`, load shedding, and the authenticated upgrade — `WsData` and `UpgradeTarget` are declared with the decision that builds them |
-| `sync-listen.ts` | binding a node to `Bun.serve` and to the shutdown hook — the only `Bun.serve` in the package |
-| `drain-evictions.ts` | evicting every socket a drain holds, in bounded chunks, and waiting out the presence leaves each eviction started |
-| `query-window.ts` | the shared pre-policy window per query id: built once, read once for N subscribers, and replaced when it is known to be wrong |
-| `client-frames.ts` | what a RECEIVED frame does to client state, and `ClientFrameTarget` — the only inbound surface the client exposes. The mirror of `sync-frames.ts` |
-| `client-harness-fixture.ts` | the injected socket + scheduler + harness both client suites drive. Excluded from the tarball |
-| `hooks-fixture.ts` | the page harness the hook and frame suites drive: a fake socket, a page reset, a fetch double. Excluded from the tarball |
-| `subscription-book.ts` | who holds which subscription, keyed by `(socket, sid)`, and the per-socket/per-tenant caps answered from it |
-| `apply-patches.ts` | folding a patch list onto a row list (`applyPatches`) or onto ids alone (`orderAfterPatches`, what a window uses) — the client's one stateless piece, and one fold, not two |
-| `type-pins.ts` | compile-time assertions `tsc` checks — the hooks answer `AsyncState`, a query ref has no server field, every handle is `Disposable` |
-| `window-lock.ts` | one FIFO lane per query id — the only thing that orders a fanout |
-| `frame-lanes.ts` | the order one socket's INBOUND frames are applied in, and the lane key each kind belongs to. `WindowLock` again, keyed differently — and it bounds no cap |
-| `live-fanout.ts` | what one change does inside one entry's lane: match, fold, one policy pass per subscriber, and the re-snapshot that repairs a desynced one |
-| `client-heartbeat.ts` | when to beat and when to give up. A policy, which is why it is not in `client.ts`'s connection lifecycle |
-| `client-contract.ts` | the client's injected shapes — `ClientSocket`, `LiveClientOptions`, `LiveHandle` — declared apart from the class that consumes them |
-| `policy-gate.ts` | the only authz seam |
-| `subscriber-gate.ts` | the per-subscriber pass of a definition's row policy, and its two counters — `rowsDenied` and `gateFailures`. Evaluates no policy of its own |
-| `live-contract.ts` | what a live query IS: `LiveQueryDefinition`, `SnapshotResult`, `LiveSubscription`. Four modules need the shape and none of them needs the registry that runs it. The **id** is not here and not anywhere in this package — it is `@ultimat3/query`'s `queryHash` |
-| `json.ts` | the wire's value types. **No hash**: the canonical form and the sharing-key hash are `@ultimat3/core`'s (`canonicalJson`, `fingerprint`), and the 32-bit `fnv1a` that stayed for `LiveCursor.digest` went with it (2026-08-24) |
-| `live-definition.ts` | the only bridge from a declared `query({ live: true })` to a registrable definition — and `policy-gate.ts`'s only caller |
-| `matcher-bridge.ts` | the only `@ultimat3/query` matcher seam — and where a patch row is narrowed to the columns the query returned |
+| `useRecord` | 13,226 |
+| `useMutation` | 21,725 (after the outbox left islands) |
+| `useQuery` | 47,696 |
+| `useChannel` | 45,585 |
+
+- **The disk boot is ONE page script (`boot.ts`, `./boot`)**, served at `/_x/page-boot/<hash>.js`
+  on a document with the scope tag and a realtime island; it wipes every other principal's stored
+  scope, then restores. `booted` reads its promise off `globalThis[Symbol.for('ultimate.page-boot')]`.
+- **An island never carries the outbox**: `boot.ts` builds it; `useMutation` and the page socket read
+  it through `outbox-slot.ts` after `page.booted`. No boot ⇒ a refused write is rejected, not queued.
+
+## Map
+
+`index.ts` / `server.ts` are the two barrels. Server side: `sync-node.ts` (+ `sync-auth`,
+`sync-frames`, `sync-upgrade`, `sync-listen`, `drain-evictions`), `channel.ts`, `presence.ts`,
+`socket.ts`, `live-query.ts` (+ `live-definition`, `query-window`, `live-fanout`, `window-lock`,
+`subscriber-gate`, `policy-gate` — the only authz seam — and `matcher-bridge`), `replicator.ts` /
+`live-replicator.ts`, the Postgres client (`pg-*.ts`, `pgoutput.ts`), the bus (`nats-*.ts`,
+`transport-env.ts`, `fanout.ts`). Client side: `client*.ts`, `socket-engine.ts` / `socket-host.ts` /
+`sync-worker.ts`, `record-*.ts`, `page-*.ts`, `boot.ts`, `offline-queue.ts`, `use-*.ts`,
+`reactivity.ts`. Shared: `sync-protocol.ts` (the wire), `cursor.ts`, `change-buffer.ts`,
+`thundering-herd.ts`, `live-contract.ts`, `json.ts` (no hash), `type-pins.ts`. Each file's header
+states its one job.
 
 ## Commands
 
@@ -999,45 +316,12 @@ bun test packages/realtime/src            # from the REPO ROOT, never from packa
 bun run typecheck
 ```
 
-**The root is not a preference.** `bunfig.toml`'s preload installs `@ultimat3/testing`'s matchers
-and Bun reads `bunfig.toml` from the cwd, so `bun test` inside this directory loads none and six
-tests fail on a missing matcher — this package's suite reading red for the shell it was run in.
-CI's `package` job spawns `bun test packages/<pkg>` with `cwd` at the root for the same reason.
-
-Changing a frame shape means adding a fixture to `sync-protocol.test.ts` — the round-trip test
-fails if a kind has no fixture — and bumping `PROTOCOL_VERSION` **when the change makes an old
-frame unreadable in either direction**. An *additive optional* field (`snapshot.entity`, 2026-08)
-is not that: `decode` builds a whitelist, so an old client drops it and a new client reads its
-absence as a defined answer. Neither is *removing a field nothing read* (`hello.resume`, 2026-08):
-the same whitelist drops an old client's copy, and a new client's omission decodes to what the
-field always held. Bumping for either refuses every in-flight client on a rolling deploy and buys
-nothing — the version guards incompatibility, not novelty. Removing a field something *does* read
-is the opposite case and bumps.
-
-**"Nothing read it" is decided by the DECODER, not by the callers — and that half is what moved
-`PROTOCOL_VERSION` to 2 (2026-08-24, BREAKING).** `hello.resume` was free because `decode` read it
-through `list()`, which answers `[]` for an absent field; `cursor.digest` and `cursor.count` were
-read through `str()` and `num()`, which **throw**. So deleting two fields no *caller* consumed
-still made the frame unreadable to a peer one deploy behind — in **both** directions, since a
-cursor rides the client's `subscribe` and the node's `snapshot`. Without the bump the skew shows up
-as a per-frame `field "digest" must be a string`, which is the same refusal with none of the
-instruction. Before claiming a removal is free, read the field's line in `decode`: a `list()` is
-free, a `str()`/`num()` is a bump.
-
-**A patch carries the result set's columns, never the table's** — `narrowRow` in
-`matcher-bridge.ts`, `As of 2026-08-20`. A `ChangeEvent` carries the whole TABLE row (that is what
-logical replication emits, and what `@ultimat3/entity`'s `setRowObserver` emits), while a live
-query's result set is whatever its `sql` returned. Every patch used to forward the change row
-unnarrowed, so a column a projection exists to withhold went out on the socket the moment it
-CHANGED — `examples/dummy`'s feed projects ten columns and one publish delivered `updatedAt` to
-every subscriber (#230). The per-subscriber gate cannot help: it decides whether a ROW is delivered,
-never which of its columns.
+`bunfig.toml`'s preload installs `@ultimat3/testing`'s matchers and Bun reads it from the cwd.
 
-`id` always survives the narrowing — it is the row's identity on the wire, and `applyToWindow` and
-every client store key by it. An **unknown** projection narrows nothing, because "nothing has been
-read yet" is not "the result set has no columns".
+Changing a frame shape means a fixture in `sync-protocol.test.ts` (the round-trip test fails
+without one) and bumping `PROTOCOL_VERSION` **when an old frame becomes unreadable in either
+direction**. `decode` is a whitelist: an additive optional field, or removing a field read through
+`list()`, is free; removing a field read through `str()`/`num()` (which throw) is a bump. Read the
+field's line in `decode` before claiming a removal is free.
 
-The projection is **learned from the query's own reads**, in `live-definition.ts`: a projection
-lives inside the `sql` provider's closure and there is nothing static to read it from. Learned and
-kept rather than re-derived per fanout, because the case the window's own rows cannot answer is an
-EMPTY window — the first row to arrive would otherwise go out whole.
+Why each rule above is shaped the way it is: [`docs/history/realtime.md`](../../docs/history/realtime.md).