Three client gaps turn transient relay failures into permanent UI
degradation. Under a slow or rate-limited relay:
1. A cold channel's profile batch exhausts its single retry and leaves
raw npubs + broken mention chips until the user manually kicks the
channel.
2. A thread opened from a notification trusts a successful-but-empty
reply read as authoritative and never retries.
3. A rate-limited `CLOSED` on a history subscription immediately rejects
the caller rather than retrying after the rate-limit window.
All three are addressed without changing global query defaults or the
happy-path behavior.
## Changes
**Fix 1 — cold profile batch resilience** (`useUsersBatchQuery`,
`desktop/src/features/profile/hooks.ts`)
Override `retry: 3` with exponential backoff and error-gated
`refetchOnWindowFocus: (query) => query.state.status === "error"`,
scoped to this query only. The global defaults (`retry: 1`,
`refetchOnWindowFocus: false`) are intentional for other queries and are
unchanged. After the retry budget exhausts, a window-focus event (e.g.
channel-switch) recovers the query automatically — but only when it is
already in an error state, preventing unnecessary refetches for
successful batches.
**Fix 2 — stale-empty thread reads** (`useThreadReplies.ts`,
`ChannelScreen.tsx`)
Add optional `expectedEventId` parameter. When a completed paged fetch
does not contain the expected event, throw
`ThreadExpectedEventMissingError` so React Query's built-in retry
machinery handles it rather than caching an authoritative empty.
`ChannelScreen` passes `threadScrollTargetId` (the notification-linked
reply ID) as `expectedEventId`.
When notification routing changes `expectedEventId` while the same
thread root is already mounted (same query key), an explicit
`invalidateQueries` in a `useEffect` triggers a fresh validation pass.
The `useEffect` is declared after `useQuery` so TanStack's internal
options-update effect installs the new `queryFn` closure first; the
refetch therefore uses the current `expectedEventId` rather than the
previous null closure. For the cold-start race (target arrives before
the first page returns), the effect detects `fetchStatus === "fetching"
&& status === "pending"` and calls
`cancelQueries().then(invalidateQueries)` so the obsolete in-flight
response cannot settle as authoritative before the new target's
validation closure is active.
The query-fn tracks consecutive fetch attempts per target. On attempt 3,
it adds the target to `exhaustedTargetsRef` before calling
`loadThreadReplies`. `loadThreadReplies` sees the target in the
exhausted set and returns the fetched replies directly rather than
throwing — the terminal attempt always resolves to success. No
re-entrant scheduling: the resolution is synchronous inside the query
function itself. Deleted/moderated targets never lock the thread in a
terminal error surface.
**Fix 3 — CLOSED recovery for history subscriptions**
(`relayClosedRecovery.ts`, `relayClientSession.ts`,
`relayClientShared.ts`, `relayGateBoundary.ts`)
On a rate-limited `CLOSED` the subscription previously rejected the
caller immediately. Store `filter` and `timeoutMs` on
`HistorySubscription`, then on rate-limited `CLOSED` re-register under a
fresh `subId` and defer `sendReq` until the rate-limit window clears —
matching the live-sub recovery design already present in
`relayClosedRecovery.ts`. Bounded to 3 attempts; exhausted retries
reject immediately so callers are never left waiting indefinitely. A new
op-timeout guards the retry REQ against a non-responding relay; when the
op-timeout fires it sends `CLOSE` for the rotated `subId` (matching the
behavior of the original timeout path) so the relay releases the slot
rather than counting it against the per-connection cap.
## Tests
- `relayClosedRecovery.test.mjs`: behavioral fake-clock tests for
history-sub retry, 3-attempt exhaustion, op-timeout CLOSE send +
late-EOSE non-regression, rejecting-`closeSubscription` swallowed
without unhandled rejection, wiring source assertion (fails if
`relayClientSession.ts` drops the `closeSubscription` callback) — 18
tests
- `useThreadReplies.test.mjs`: `loadThreadReplies` unit tests
(throw/exhaustion-guard); behavioral hook tests via real
`QueryClientProvider` + `renderHook`: exhaustion-resolves-to-data,
settled null→target change retries on missing-target page and lands
target data (fails if `invalidateQueries` is removed OR if `useQuery` is
moved after the `useEffect`), cold-fetch cancel-then-invalidate (gated
fetcher — released after rerender, stale empty discarded, replacement
fetch settles with target); ChannelScreen wiring source assertion — 9
tests
- `profileBatchResilience.test.mjs`: source assertions for `retry: 3`,
`retryDelay`, error-gated `refetchOnWindowFocus`, and unchanged global
defaults — 2 tests
---------
Signed-off-by: Will Pfleger <pfleger.will@gmail.com>
Co-authored-by: Duncan <dcfd242e557282d7a1e2cf2e6877522682f1e5c6156dc92ca7d90eaedd3b0f95@buzz.block.builderlab.xyz>
Pinky, an AI agent, updated this description on Wes's behalf after
taking over the startup investigation.
**Category:** fix
**User Impact:** An EVENT refused by WebSocket admission or handler
saturation receives a correlated `OK(event_id, false, reason)` instead
of an uncorrelated NOTICE, so the client can settle that refusal without
waiting for its publish timeout. Rate-limited refusals also arm client
backoff. This fixes a protocol failure mechanism; it does not establish
that every startup send will succeed or that the reported Desktop
startup incident is fully resolved.
**Problem:** Startup opens several live subscriptions and publishes at
once, and the relay's WebSocket admission gate is a fixed 5-second
window (`ws_admission_budget` = `human_ws_events_per_sec * 5`). If that
shared per-principal quota is exhausted, `enforce_ws_admission`
previously rejected an EVENT with a bare `["NOTICE", reason]`. Quota
pressure is a possible trigger, not proof of the original incident's
complete cause.
A NOTICE carries no event id. Both clients settle a pending publish
*only* from an `OK` keyed by event id (desktop `pendingEvents`, mobile
`_pendingEvents`), so nothing settled — and `handle_text_message`
returns early, so no `OK` ever followed either. The send **could not
fail**; it could only time out at `PUBLISH_TIMEOUT_MS` = 25s. That
explains how this rejection mechanism can produce a roughly 25-second
timeout; attributing the original report to it still requires the actual
startup/send workflow.
The handler-semaphore saturation path had the identical defect, and that
one needs no quota burst to fire.
**Solution:** NIP-01 gives each request type its own acknowledgement
channel, and a rejection is only actionable on the same one. Reject a
REQ with `CLOSED`, an EVENT with `OK(id, false, reason)`, and fall back
to `NOTICE` only where no per-request correlation exists. COUNT refusals
now also use `CLOSED(query_id, reason)` per NIP-45, covering both quota
admission and handler saturation (added in
`cd12c93804b87a24b61075dfd171dc471a0a527f`).
Reason strings are unchanged, so the `rate-limited:` prefix and `retry
in {N}s` hint that existing client gates parse keep working (desktop
`parseRateLimitHint`, mobile `RelayRateLimitGate`, buzz-acp
`set_rate_limit_gate`). Only the frame *type* changes, so
`docs/multi-tenant-relay.md` L7 stays satisfied.
Two notes on how this landed, both worth a reviewer's attention:
1. **A survived mutation became a design change.**
`send_admission_result` originally took a `RejectionTarget` parameter,
and reverting the *second* call site (the per-minute message quota)
survived the whole suite — with Redis unreachable the first quota check
short-circuits, so that line is unreachable in test. Rather than test
around it, the parameter is gone: the target is derived from the frame,
so no call site can name the wrong channel.
2. **The relay fix would have caused a client regression on its own.**
Gate arming lived only in the NOTICE branch. Once rejections arrive as
`OK:false`, `handleOk` failed the send without ever backing off — the
client would retry straight into the same quota. Desktop and Mobile now
arm on a `rate-limited:` OK rejection. ACP was subsequently fixed in
`3b06dd32493596ec650f20abf8805791c50fdc24`: it arms the gate and
re-parks only the refused observer frame, preserving other in-flight
frames. Desktop gets `activateRateLimitIfSignalled` as the single owner
of that prefix test, called from both `handleOk` and the NOTICE branch.
<details>
<summary>File changes</summary>
**crates/buzz-relay/src/rejection.rs** (new)
Owns the admission-rejection concern: `RejectionTarget`,
`rejection_target_for`, `request_rejection_message`,
`send_admission_result`, and `enforce_ws_admission`, moved out of
`connection.rs`. Six tests, two of which drive the real
`enforce_ws_admission` against a real `AppState`.
**crates/buzz-relay/src/connection.rs**
Fix the EVENT handler-semaphore rejection to correlate to the event id;
delegate admission to the new module. Add two tests that drive the real
`handle_text_message` with every handler permit held. Down from 1319 to
1116 lines.
**crates/buzz-relay/src/state.rs**
Widen the existing `test_state` helper to `pub(crate)` so the rejection
tests reuse it rather than adding a ninth copy of `AppState`
construction.
**desktop/src/shared/api/relayRateLimitGate.ts**
Add `activateRateLimitIfSignalled` — one owner for the `rate-limited:`
prefix test, since three inbound frame types now carry it.
**desktop/src/shared/api/relayClientSession.ts**
Arm the gate on a rate-limited OK rejection; route the NOTICE branch
through the same helper. Net zero lines, which keeps this
already-oversized file within the differential ratchet.
**desktop/src/shared/api/relayClientPublishRejection.test.mjs** (new)
Four tests against the real `RelayClient`: a rate-limited OK settles the
pending publish and arms the gate; an ordinary rejection does not arm
it; an accepted OK still resolves.
**mobile/lib/shared/relay/relay_session.dart**
Arm the gate in `_handleOk` for a rate-limited rejection.
**mobile/test/shared/relay/relay_session_test.dart**
Two tests driving the real `publish` + `debugHandleMessage` path.
</details>
<details>
<summary>Validation</summary>
**Mutation-tested — 5 mutations, all now killed.** Each production call
site was reverted to the defective behaviour to confirm a test fails.
This caught two false-negative tests:
| # | Mutation | Result |
|---|----------|--------|
| 1 | `rejection_target_for`: EVENT → `Connection` | 4 tests fail |
| 2 | EVENT handler-semaphore call site → bare NOTICE | **survived at
first** |
| 3 | per-minute quota call site → `Connection` | **survived**; fixed by
removing the parameter |
| 4 | desktop `handleOk` gate arming removed | 1 test fails |
| 5 | mobile `_handleOk` gate arming removed | 1 test fails |
Mutation 2 is the lesson: my first saturation test called
`request_rejection_message` directly, so reverting the real call site
inside the `match` arm left it green. It now drives
`handle_text_message` itself and dies on that mutation.
- `cargo test -p buzz-relay` — 928 passed, 1 failed:
`api::mesh_demo::tests::demo_join_forwarded_arm_round_trips_echo`,
**pre-existing**, reproduced with all changes stashed at `4dd4d73de`.
- `cd desktop && npm test` — 5721 passed, 0 failed (full suite).
- `cd mobile && flutter test` — 1876 passed, 0 failed (full suite).
- `just fmt-check`, `just clippy`, `just desktop-check`, `just
mobile-check`, `just file-size-check` — clean. Desktop's 5 biome
warnings are pre-existing (reproduced with changes stashed).
- All 9 pre-push lanes green, including `rust-tests` and
`desktop-tauri-checks`.
**Not verified:** not reproduced end-to-end against a live relay under a
forced quota burst. The causal chain is source-proven and
mutation-proven at the frame level; the ~25s attribution follows from
`PUBLISH_TIMEOUT_MS` but is not directly measured. A packaged-build
click-through would close that gap.
</details>
Related work: #6957 bounds Desktop HTTP event submission, but safe
retained-operation recovery after exhausted/ambiguous outcomes remains
unfinished. #6998 is the separately reviewable Desktop
readiness/duplicate-subscription slice. Neither is claimed to complete
native before/after startup-send validation.
Diagnosis note: `RESEARCH/DESKTOP_STARTUP_SEND_STALL_2026_08_27.md`
(Brain's workspace).
## Current review disposition (2026-08-28)
The [review on
`cd12c938`](https://github.com/block/buzz/pull/6961#pullrequestreview-5052902510)
identified ACP's missing rate-limited-OK handling. Commit
`3b06dd32493596ec650f20abf8805791c50fdc24` fixes gate arming, re-parking
the specifically refused observer frame, and the stale NOTICE comment.
Two regressions drive the real frame dispatcher. See [the implementation
and validation
response](https://github.com/block/buzz/pull/6961#issuecomment-5455032054).
The Mobile generation-check inline thread is resolved: its `async
publish` returns a failed Future when superseded; it does not throw
synchronously at invocation. No further production change was indicated
by that comment.
The validation counts above describe the original slice, not a new
rerun. At `3b06dd324`, the current GitHub check rollup has successful
completed test/build checks (non-applicable jobs skipped). The
security-review comment still requires review for the current base/head
range; do not read a green authorization job as a completed security
review. Approval and merge remain human decisions.
---------
Signed-off-by: Brain <1a02c72794dcd0f07058a353bc3a81f4028b8c77c92c87fce6d5c8b85970a20b@buzz.block.builderlab.xyz>
Signed-off-by: Wes <wesbillman@users.noreply.github.com>
Co-authored-by: Brain <1a02c72794dcd0f07058a353bc3a81f4028b8c77c92c87fce6d5c8b85970a20b@buzz.block.builderlab.xyz>
Co-authored-by: Carl <9d00794d3df50972eb8b615511783cab12a77a8fd5dd5edd58073ec73b54bd8b@buzz.block.builderlab.xyz>
## Summary
- repair exact-channel reconnect gaps from a source-coupled 1,865-second
floor (`900` seconds of relay future tolerance + `960` seconds of DB
created-at floor + `5` seconds of fence margin)
- page the existing authenticated `POST /query` bridge losslessly with
`(until, before_id)`, including dense timestamp boundaries and an
unbounded first page that does not trust the renderer clock
- keep the restored live REQ unchanged while deduplicating live/repair
overlap until both live EOSE and repair completion for the active
connection generation
## Scope and behavior
The wider repair applies only to live subscriptions containing the exact
channel-event kind set and one `#h`. Profile, read-state, global,
channel-less, and other subscription behavior is unchanged. Failed or
stale repairs retain their pinned lower bound for the next reconnect.
This intersects the replay path described in #3215, but does not claim
to fix that issue. Desktop already has a bounded notification-side
event-ID guard; this PR adds transport dispatch dedupe for the
restored-live/repair overlap. Notification behavior should remain a
focused regression check while this wider repair window rolls out.
## Testing
- `cd desktop && pnpm typecheck`
- `cd desktop && pnpm biome check
src/shared/api/channelReconnectRepair.ts
src/shared/api/relayClientSession.ts src/shared/api/relayClientShared.ts
src/shared/api/relayClosedRecovery.ts
src/shared/api/relayReconnectReplay.ts
src/shared/api/relayReconnectReplay.test.mjs src/testing/e2eBridge.ts`
- `cd desktop && node --test
src/shared/api/relayReconnectReplay.test.mjs` (26 passed)
- `cargo test --manifest-path desktop/src-tauri/Cargo.toml
channel_reconnect_repair` (2 passed)
- pre-push hooks passed at `743283439eb6be7969e19951cbb6720c5e493a04`
---------
Signed-off-by: Wes <wesbillman@users.noreply.github.com>
Co-authored-by: Carl <32a2e2c9d428ee08902cab75d956da2c1d235a22d4766b0dd4138bf6e2e5db1d@buzz.block.builderlab.xyz>
## Summary
Moves five hot renderer paths out of JavaScript and into Rust, targeting
main-thread CPU on the desktop app: wakeups, IPC round trips, and
per-frame JS work during reconnect storms, catch-up, and archive sync.
Five commits, reviewed and ratified sequentially (each round included
independent re-review, mutation-tested coverage, and full-suite gates at
pinned heads):
| Commit | Change |
|---|---|
| `9b5a82ab9` | Batch inbound relay frames in the native websocket
plugin: N run-loop wakeups collapse into 1 per 8ms window, byte-bounded
so no batch crosses the Tauri direct-execute threshold onto the slow
fetch path. NIP-42 AUTH bypasses the window. |
| `8b047a35e` | Move the local-archive subscription into Rust: the
renderer no longer sees archived frames at all. Shared
`native_relay_client` (one authenticated socket per relay+pubkey,
declarative subscription reconciliation, CLOSED-driven retry with
write-time eviction), lease/epoch lifecycle ownership so delayed IPC
cannot resurrect a stopped task. |
| `609962b49` | Move persona catalog fetching into Rust: signature
verification (~473us/event, ~0.24s per 500-event page formerly on the
webview thread) now runs under `spawn_blocking`; one projected DTO
crosses IPC. |
| `38afea0a0` | Batch unread catch-up behind one IPC call: N per-channel
REQs/awaits become one command with global two-pass classification,
making the completion-order race unrepresentable. Both IPC DTO surfaces
pinned by wire tests asserting serde output against the TS contract. |
| `b955f528b` | Move observed unread state into native SQLite: scoped
WAL DB with a sequence/revision protocol, transactional localStorage
migration, epoch-tagged rebuild detection. Membership no longer
serializes five capped arrays per catch-up: measured 327.4 KiB before,
0.2 KiB after at the same populated 5x1000 fixture. |
Full design rationale, failure matrices, and verification detail are in
the individual commit messages.
### Known behavior notes
- **Native migration is a one-way door per profile.** A successful
native open sets `migration_complete` in the first open's transaction
and the renderer then deletes the legacy localStorage key
(`useObservedUnreadPersistence.ts:290-294`,
`observed_unread.rs:420-442`). The one-release JS fallback remains for
native open *failure*, but a profile that has completed migration cannot
return to the JS store. Relevant if a rollback path is ever needed.
- **The membership seed is built unconditionally at startup**
(`useUnreadChannels.ts:245-250`), before any `isNative()` check. It is
ref-cached and small, but it is a startup cost the baseline does not
pay, including on the fallback path.
### Related issue
None found. Work originated and was reviewed in the Buzz
`buzz-gui-performance` channel.
### Testing
**Current head is `9a8128ba4d2d695113505162be388fcec62cda0a`. For
receipts at that head, see [Round-3
receipts](#round-3--final-head-9a8128ba4) at the bottom; the block
immediately below is the historical record for the original five-commit
pack and its counts are superseded.**
The receipts in this subsection are at exact pack head
`b955f528b6dfa34db089b21cc718fba0f9284d6f` (tree oid `dcc711a66`),
verified by three independent instruments (implementer, reviewer,
gatekeeper), with `git rev-parse HEAD` confirmed in the same shell as
each run:
- `cargo test --lib` (full package): **2511 passed / 0 failed** (+7 +3
in the other targets)
- `pnpm test`: **4920 / 4920 across 74 suites**
- `cargo fmt --check`, `clippy --all-targets -D warnings`, `tsc
--noEmit`, `pnpm check` (including the file-size ratchet): all clean
- Mutation controls on the final review round: badge-lane controls A and
B each turn exactly one test red; seed-guard (M4) and
anchor-monotonicity (M5) mutants each killed by exactly their own
witness; restored bytes fully green
- The observed-unread native path is exercised by an in-tree
`__TAURI_INTERNALS__` protocol rig (entry, replay, gap, rebuild,
markers, membership, badge lane), added after review found the original
suite never left the localStorage fallback
**Open gates, stated plainly (status as of head `9a8128ba4`):**
- **Paired CPU benchmark still not run.** The three-arm design (baseline
`78cbffeb6` vs pack vs a null arm with native forced off, fresh profile
per arm) was instrument-reviewed but never executed. The IPC payload
reduction above is measured; end-to-end CPU improvement is predicted,
not measured. This gate is unchanged since `b955f528b`.
- **Live-local gate at the final head: GREEN at `9a8128ba4`.** A real
native Tauri live-local pass per `TESTING.md` was run at this exact
head. Receipts:
`.scratch/pr6024-final-9a8128ba4d2d695113505162be388fcec62cda0a/`,
`MANIFEST.txt` `start_utc=2026-08-17T21:58:07Z`, `FINAL_ATTESTATION.txt`
`end_utc=2026-08-17T22:20:47Z` — both after the commit's `21:32:51Z`,
and the build worktree's HEAD is `9a8128ba4` with a clean tree. **I
opened these files rather than relaying the summary**, and recomputed 6
SHA256SUMS entries against the sealed manifest (6/6 match). From the raw
artifacts: historical catch-up persisted 10 seeded events into the
native `observed-unread.db`; a realtime event advanced it 10→11; the
relay was killed and restarted with Desktop up and an event sent after
recovery advanced it 11→12 within 2 polling ticks; Desktop was then
stopped, an event sent offline, and after relaunch with the appdir
preserved catch-up advanced it 12→13. `relay.log` shows 2 app WebSocket
connections, the restart-close signal, and fresh connections after each
recovery — the app-attributable socket evidence the previous round
lacked. The preserved appdir database is still on disk and `select
count(*) from observed_events` returns **13**. The ignored real-relay
archive wire test passed **1/1** at this head. This exercises the
`lib.rs` / `e2eBridge.ts` merge recombination under a real app and
blocker 2's failure-mode family (relay killed mid-session, app killed
with offline traffic in flight, sync did not go dead). It does **not**
directly force the precise overlapping B/C archive-ownership timing
window through the GUI; that invariant is pinned by the compiled
`drop(owner)` mutation proof below, and is deliberately not relabelled
as live race evidence. **Retraction retained for the record:** an
earlier revision of this description said live-local was GREEN at
`9a8128ba4` while citing a run executed against `f079d0914`, the pre-fix
head, roughly three hours before `9a8128ba4` existed. That claim was
false and was corrected before this run existed. The GREEN above rests
on the new receipts only.
- **Workspace-wide `cargo test --workspace` is red on `main`, not on
this PR.** The exact-head live run's workspace logs contain honest
`rc=101` results. All are in crates this PR does not touch: `git diff
origin/main...9a8128ba4` is **empty** for `crates/git-sign-nostr`,
`crates/buzz-relay`, and `crates/buzz-pair-relay`.
`git-sign-nostr::tests::test_parse_envelope_rejects_invalid_oa_pubkey`
fails **deterministically on clean `main`** — reproduced independently
in fresh worktrees at merge-base `a282e0643` and at current tip
`7f61cf431` (55 passed / 1 failed, same assertion at `lib.rs:2136`).
Root cause: `nostr` 0.36→0.44 changed `PublicKey::from_hex` from a
curve-point parse to a plain hex decode, so an all-zeros (off-curve) key
now parses; measured with a standalone probe, `from_hex("0"*64).is_ok()`
is `false` on 0.36.0 and `true` on 0.44.7. Filed as block/buzz#6175,
with the observation that nothing in CI runs this crate's tests (`just
test-unit` enumerates packages and omits it; `server-cross-compile`
compiles it and runs nothing). The mesh-demo and pair-relay failures in
the same logs are timing flakes in equally untouched crates. **None of
these gate this PR.**
- **Full smoke + integration: GREEN at `9a8128ba4`.** CI run 32072147445
is `success` with every job green — all 4 Desktop Smoke E2E shards, both
Desktop E2E Integration shards, Desktop Core, Desktop Build (macOS),
Desktop E2E Relay, Rust Lint, Windows Rust, DCO, Dead Token Reference
Guard. Attempt 1 was cancelled by an infrastructure hang, not by this
code: both integration shards stalled 19m37s inside `Install Playwright
system dependencies` (an apt install that runs before the e2e bundle is
built or the relay starts), so no test executed and all downstream steps
were skipped. The same step took 16s at `9128b9389` and 12s/18s on
attempt 2 of these same bytes. Attempt 2 was a job rerun, not a push;
the head is unchanged.
---
Opened by Eva (agent) on Tyler's behalf at his request.
---
## Review round 2 — head `9128b9389`
Carl's review
[4953970089](https://github.com/block/buzz/pull/6024#pullrequestreview-4953970089)
raised two blockers against `f079d0914`. Both were confirmed real at the
source by a second reader and both are fixed in `9128b9389`.
**Blocker 1 — recovery dropped the mutation whose ingest failed.** Every
captured mutation now gets one retry after `reopen()` refreshes
sequence/revision, which is safe because ingest is idempotent (events
upsert `DO NOTHING`, channel latest advances by `MAX`, membership is
`INSERT OR IGNORE`, and a replayed sequence returns a snapshot rather
than reapplying). `snapshotRequired` takes the same recovery path
instead of being treated as success. If reopen and retry both fail,
native is explicitly degraded and equivalent state is applied to the JS
fallback so `isNative()` stops reporting healthy. Regressions assert the
*originally rejected* marker, destructive clear, and membership delta
each survive; removing the retry takes 5 tests red including all three.
**Blocker 2 — a superseded archive start could tear down the newer
scope's relay session.** `begin` now returns an `ArchiveOwnership` token
that holds both ownership guards, and `archive_session` requires one, so
session acquisition happens inside the ownership critical section rather
than after it.
### Deviation from the requested regression test for blocker 2
Carl asked for "a production-wiring regression that pauses B after
`begin`, fully installs C, resumes B, and proves only C's session and
subscriptions remain live." **We did not write that test, deliberately,
and wrote a stronger one instead.**
Under this fix that scenario is unreachable by construction. The fix is
not a revalidation check that a test could observe passing or failing —
it is a type-level fence: `begin` hands the winner a token that holds
the ownership locks, and `archive_session` cannot be called without one.
A test that staged "B pauses after `begin`, C fully installs" would be
staging a state the type system now forbids, so it could only pass
vacuously.
Two things close the blocker, and only one of them is a test:
1. *A superseded start cannot reach `archive_session` at all.* Enforced
by the compiler, not by a test. `ArchiveOwnership` is un-constructible
outside `archive::sync`. Verified by attempting both bypasses: calling
`archive_session` without a token fails **E0061**, and forging the token
from a real non-archive caller fails **E0451** ("fields `_latest` and
`_running` are private"), with a second independent barrier because
`RunningSync` is itself a private type.
2. *While an owner holds the token, no newer start can claim.* This is
the property the token's usefulness rests on, and it is what the new
regression pins —
`a_newer_start_cannot_claim_while_the_owner_holds_its_token`.
That test is aimed at the one mutant that actually threatens this
design: keeping the token but releasing the guards inside `begin`, which
is what someone reaches for to avoid holding a lock across an await. It
**compiles**, it keeps every other lifecycle test green, and it restores
the exact race. Against it the new test goes **single-red** — the other
26 tests in the module stay green.
Proving mutual exclusion over the whole claim → acquire → install unit
is strictly stronger than staging a pause the type system forbids, which
is why we chose it.
Also recorded in the code, in `ArchiveOwnership`'s doc comment: what
this token does **not** cover. It serializes archive lifecycle against
archive lifecycle, not against a plain `NativeRelayClient::session()`
call from a non-archive feature. That window predates this PR and is
theoretical today given renderer gating, so it is documented rather than
fixed here.
### Round-2 receipts at `9128b9389`
- `cargo test --lib`: **2513 passed / 0 failed**
- `pnpm test`: **4926 / 4926**
- `cargo clippy --all-targets -D warnings`, `cargo fmt --check`, `pnpm
check`: rc 0
- Merged against `origin/main` in a scratch worktree (clean, zero
conflicts) and re-gated on that merged tree: **2544** Rust / **4956**
frontend, both 0 failed
---
## Round 3 — final head `9a8128ba4`
`9a8128ba4d2d695113505162be388fcec62cda0a` is a merge of `origin/main`
(`a282e0643`) into the pack branch, taken **before** final verification
so that verification would run on the bytes CI builds rather than on a
head that was about to move.
The merge is textually clean but touches two files the pack also
touches, `desktop/src-tauri/src/lib.rs` and
`desktop/src/testing/e2eBridge.ts`, so recombination was checked
explicitly rather than inferred from the absence of conflicts: **all 342
pack-added lines in those files survive (0 missing), and all 17
main-added lines survive (0 missing)** — no silent revert in either
direction. Main's contribution is a `#[doc(hidden)] pub use
print_agent_access_owner_only_probe_if_requested` in `lib.rs`, plus a
Bumble→Pollen rename and workflow revision / `workflowUpdateError` mocks
in `e2eBridge.ts`.
### Receipts at `9a8128ba4`
Implementer, in a shell with `git rev-parse HEAD` confirmed at this SHA:
- `cargo test --lib` (full package): **2544 passed / 0 failed** / 18
ignored
- `pnpm test`: **4956 / 4956** across 74 suites
- `cargo clippy --all-targets -D warnings`, `cargo fmt --check`, `pnpm
check`: rc 0
- The 4 remaining `pnpm check` findings (`channelMutesStorage.test.mjs`,
`channelStarsStorage.test.mjs`, `terminal.css`,
`empty-edit-delete.spec.ts`) are all in files this PR never touches
- All 7 pre-push hooks green, including `branch-skew`, which now passes
on its own because the branch contains `origin/main`
Independent reviewer, same exact head, own instruments:
- `cargo test --workspace`: green; `cargo clippy --workspace
--all-targets -- -D warnings`: green *(a later workspace run at
`9a8128ba4` hit pre-existing `main`-side failures in untouched crates —
see the workspace-gate entry under Open gates above; the difference is
flake timing plus which run happened to reach `git-sign-nostr`, not a
change in this PR)*
- `pnpm test`: **4956 / 4956**; `pnpm check`: rc 0, same known findings
outside touched files
- **Mutation proof of the blocker-2 fix:** dropping the ownership
token's hold (`drop(owner)`) turns
`a_newer_start_cannot_claim_while_the_owner_holds_its_token` red;
restoring the shipped bytes returns the suite green. The invariant that
a superseded start cannot claim or disturb a newer archive session is
pinned by a test, not just by the type system.
- **Live-local at this exact head: attested and verified.** The
mis-cited `f079d0914` run that originally appeared here is retracted; a
real run at `9a8128ba4` replaced it, and its receipts were opened
file-by-file (not relayed) by a second agent. See the live-local entry
under Open gates above for the receipt path, wall-clock bounds,
store-count chain, and socket evidence.
- Verdict: minimalness **9/10**, elegance **9/10**, correctness
**9.5/10**. **Score restored** now that the exact-head live-local run
exists and its provenance has been checked from the files (SHA-named
receipt dir, self-dating manifest, receipt mtimes after the commit time,
sealed digests spot-checked). Gatekeeper re-bless at `9a8128ba4`:
**9/10, merge recommended.**
The paired CPU benchmark remains the one gate not run. It is a
measurement of the pack's *benefit*, not of its correctness, and it is
stated as unrun rather than estimated.
---------
Signed-off-by: Tyler Longwell <tlongwell@squareup.com>
Signed-off-by: Wren <5217c5c2f7bfb4333e46d17c98a9255a52dadee18dcd43a43536b95e6776dfa0@buzz.block.builderlab.xyz>
Signed-off-by: Eva <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz>
Signed-off-by: Wes <wesbillman@users.noreply.github.com>
Co-authored-by: Dawn <c6237ef84fa537c78dcee78efd2d4e59f728859c7f194da42ac51ededfa0be05@buzz.block.builderlab.xyz>
Co-authored-by: Wren <5217c5c2f7bfb4333e46d17c98a9255a52dadee18dcd43a43536b95e6776dfa0@buzz.block.builderlab.xyz>
Co-authored-by: Perci <5a968df9a7494b4e019b9ecf739e088ba61097b4312124e9a88ae5b42e3f5f3e@buzz.block.builderlab.xyz>
Co-authored-by: Eva <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz>
Co-authored-by: Wes <wesbillman@users.noreply.github.com>
Co-authored-by: Carl <c7ebe626f000404285d3686e1dc74cc07cc60a9754a150041ba132e14bd3e2ec@buzz.block.builderlab.xyz>
## Summary
- buffer native WebSocket frames until `plugin:websocket|connect`
returns the socket ID and the NIP-42 auth waiter is installed
- drain those frames in order before normal inbound handling begins
- add a deterministic E2E seam and regression for an AUTH challenge
delivered before connect resolves
## Root cause
The native WebSocket task starts forwarding relay frames before the
connect command returns its socket ID. Buzz relay sends the NIP-42 AUTH
challenge immediately on connection, so Desktop could process that
challenge while `wsId` and `authRequest` were still unset. The challenge
was discarded, the cold connection failed authentication, and the first
plain-text send failed. Channel history still appeared because it loads
through the Tauri channel-window command rather than this frontend
WebSocket.
## Verification
- deterministic regression is red before the fix (`connecting` after 5s)
and green after it
- `pnpm test`: 3,721 passed
- `pnpm build:e2e`: passed
- focused early-AUTH E2E: passed
- existing failed-initial-dial E2E: passed
- `pnpm typecheck`: passed
- Biome + file-size pre-commit checks: passed
The complete 7-test relay-reconnect file passed once before the final
line-count-only compaction. After that compaction, two full-file reruns
each had the same unrelated startup-seam flake in the existing
initial-dial test; both the new test and that existing test pass
independently at the committed HEAD.
---------
Signed-off-by: npub1jh9wn95s0472h86ahapupaf7m6kx4v9sx2n0atj2hltcfer8k06s5n3pyf <95cae996907d7cab9f5dbf43c0f53edeac6ab0b032a6feae4abfd784e467b3f5@buzz.block.builderlab.xyz>
Signed-off-by: Wren <5217c5c2f7bfb4333e46d17c98a9255a52dadee18dcd43a43536b95e6776dfa0@buzz.block.builderlab.xyz>
Co-authored-by: npub1jh9wn95s0472h86ahapupaf7m6kx4v9sx2n0atj2hltcfer8k06s5n3pyf <95cae996907d7cab9f5dbf43c0f53edeac6ab0b032a6feae4abfd784e467b3f5@buzz.block.builderlab.xyz>
Co-authored-by: Wren <5217c5c2f7bfb4333e46d17c98a9255a52dadee18dcd43a43536b95e6776dfa0@buzz.block.builderlab.xyz>
## Summary
- replace the desktop's global kind-20001 presence subscription with one
author-filtered subscription derived from active TanStack presence
queries
- reconcile changing demand without a delivery gap: promote only after
relay EOSE, keep the last confirmed subscription on failure, discard
stale opens, and close entirely when demand is empty
- preserve REST presence as the initial seed and TTL/crash-recovery
backstop
- add transport-seam and lifecycle tests for readiness, normalization,
churn, retries, close failures, reconnect ownership assumptions, and
disposal
## Why
The desktop currently receives presence heartbeats from every identity
on the relay. A live tap measured roughly 2,700 events/minute (45/sec),
about 1 MB/minute and 71.5% of readable traffic, from approximately
1,300 distinct fleet identities. Most are discarded only after
WebSocket, Tauri IPC, and JS parsing.
This change applies normal Nostr author filtering at relay fan-out,
before those costs. It deliberately does not introduce a relay digest
protocol or client-side event batching; relevant-author traffic should
be small after scoping, and the existing signed-delta/REST-TTL model
remains intact.
## Correctness model
- active query observers are the demand source; inactive cached queries
retain no authors
- replacement opens before old closes and is promoted only after EOSE
- timeout/CLOSED rejects and closes the candidate while preserving the
last good subscription
- rapid A→B→C and A→B→A churn cannot unseat current A with stale B
- empty demand never sends an unfiltered subscription
- RelayClient continues to own reconnect replay; the reconciler does not
duplicate subscriptions on reconnect
## Validation
Exact pushed head: `8845093aec0330be16efe52d3459ff67f1000ff4`
Pre-push hooks passed:
- desktop check and file-size ratchet
- desktop TypeScript
- desktop unit suite: 4,791/4,791
- branch-skew check
Focused lifecycle/transport suite: 34/34 passed before commit.
Independent Royal Court review found and blocked two prototype flaws
(timeout-as-success and starvation-prone trailing debounce); both were
fixed and the final worktree was cleared with no remaining correctness
or lifecycle blockers.
Signed-off-by: Wes <wesbillman@users.noreply.github.com>
Co-authored-by: Carl <c7ebe626f000404285d3686e1dc74cc07cc60a9754a150041ba132e14bd3e2ec@buzz.block.builderlab.xyz>
**Category:** fix
**User Impact:** Previously selected community themes are preserved when
opening a relay through onboarding, while first-time theme migration
still completes for communities with no saved theme.
**Problem:** During community initialization, desktop queried theme
history before establishing live delivery. If a replacement event
arrived while an empty history query was in flight, the client could
incorrectly treat the theme as absent and publish the default over the
user's saved selection.
**Solution:** Subscribe before fetching history, expose whether live
readiness reached EOSE, flush buffered live events before resolving
EOSE, and retain the newest delivered replacement through hydration.
Seed the inherited/default theme only when both live and history
snapshots reach EOSE with no valid or unreadable event; subscription
failures, CLOSED, readiness timeout, relay failure, and unreadable
events fail closed without publishing.
<details>
<summary>File changes</summary>
**desktop/src/shared/api/relayClientSession.ts / relayClientShared.ts /
relayClosedRecovery.ts**
Distinguish EOSE from CLOSED/timeout readiness and flush buffered events
before resolving an EOSE fence.
**desktop/src/shared/theme/CommunityThemeController.tsx**
Seed and complete first-community migration only for confirmed absence;
uncertain hydration remains non-publishing.
**desktop/src/shared/theme/communityThemePreference.ts**
Keep the inherited appearance for the first migrated community and the
stable default for later empty communities.
**desktop/src/shared/theme/communityThemeSync.ts**
Arbitrate live and history results into valid, confirmed-absent,
invalid, or unavailable hydration outcomes.
**Tests**
Cover EOSE/CLOSED readiness, subscription failure, timeout, unreadable
and live-racing events, no-op initialization, and first-to-later
community fallback isolation.
</details>
## Reproduction steps
1. Save a non-default appearance for a community relay.
2. Remove the community locally, then open the same relay again through
onboarding.
3. Arrange for the saved replacement event to arrive live while the
initial history query returns empty.
4. Confirm the saved appearance remains selected and the client does not
publish the default theme over it.
5. On an account with no theme records, open a first empty community and
confirm its inherited appearance is migrated; open a later empty
community and confirm it starts from the stable default.
## Validation
- Pre-push desktop checks, typecheck, and full desktop tests: passed at
`f79556b0e`
- Focused theme/relay readiness tests: 38 passed
- Desktop file-size ratchet and diff check: passed
---------
Signed-off-by: Taylor Ho <taylorkmho@gmail.com>
Signed-off-by: Carl <acda9e433d19dcd0e6b6840f7f4b98f3a56f1fab98049d444c087019e6d36560@buzz.block.builderlab.xyz>
Co-authored-by: Carl <acda9e433d19dcd0e6b6840f7f4b98f3a56f1fab98049d444c087019e6d36560@buzz.block.builderlab.xyz>
> Opened by Brain (agent) on behalf of @wesbillman.
## Problem
Users report the desktop app doesn't reliably reconnect and can wedge in
states where only CMD+R (or a full restart) restores connectivity
(thread `c2205e2b` in #desktop-reconnecting).
Pinky's empirical light-switch matrix (real `buzz-relay`, SIGTERM/1012 +
SIGKILL × 1s/45s/3min, at `f18a9cb10`) passed 4/4 — the backoff state
machine recovers cleanly from ordinary relay loss. That isolates the
user-stuck states to four special cases a reload resets but the auto
flow never did.
## Fixes
| Gap | Change |
|---|---|
| **G1** — recovery rode solely on the backoff timer (max 30s),
throttled by WKWebView in occluded/background windows; nothing fired on
network return or wake | New `useRelayResumeTriggers`: `online`, window
focus, and visibility→visible call `preconnect()` when the session is
`reconnecting`/`stalled`, rate-limited to one attempt per 5s
(`relayResumeTriggerPolicy.ts`). Deliberately inert for the terminal
`disconnected` state. |
| **G2** — any AUTH `OK false` latched the session terminal forever,
though the relay also rejects for transient causes (duplicate-AUTH
"already authenticated" race, ±60s clock skew, fail-closed allowlist DB
errors) | New `AuthOkTracker` (`relayAuthPolicy.ts`): "already
authenticated" resolves as success; transient rejections retry with
normal backoff; latch only on `restricted:` or after 3 consecutive
rejections. |
| **G3** — an `auth-required:` CLOSED (REQ racing AUTH after reconnect)
permanently deleted the live subscription with no UI signal — frozen
channel while state reads "connected" | Reclassified `auth-required:` as
retryable in `relayClosedPolicy.ts`. Genuinely terminal classes
(`restricted:`, `invalid:`, …) still delete. Can't loop: a truly
unauthenticated session latches terminal at the connection level. |
| **G4** — `useRelayAutoHeal` observed the 2s-debounced connection hook,
so sub-2s flaps never triggered the heal even though `resetConnection`
had already rejected every in-flight query | Auto-heal now observes the
raw connection-state emitter. The existing 15s heal rate-limit still
guards against flap storms. |
Each fix is a colocated pure-policy module + unit tests, matching the
existing `relayReconnectPolicy`/`relayClosedPolicy` pattern.
## Validation
- Full desktop unit suite: **4151 pass, 0 fail** (at branch tip, `pnpm
-C desktop test`)
- `pnpm -C desktop typecheck` and `pnpm -C desktop check` clean
(file-size ratchet respected — `relayClientSession.ts` net −2 lines
despite the tracker wiring)
- Evidence trail: `RESEARCH/DESKTOP_RECONNECT_CMDR_GAP_AUDIT.md`
(audit), `RESEARCH/DESKTOP_RECONNECT_LIGHT_SWITCH_RESULTS.md` (Pinky's
matrix)
## Not covered / follow-ups
- Native macOS sleep-wake was not automated (would kill the harness
session); G1's focus trigger is the mechanism that covers wake in
practice, but a manual sleep-wake verification on a real build is
worthwhile.
- G3 terminal-CLOSED classes (`restricted:` etc.) still silently delete
subs with no UI signal — surfacing that is a separate UX decision.
- Stall-watchdog latency (60s idle + 10s check) left unchanged; G1
triggers largely mask it.
---------
Signed-off-by: Wes <wesbillman@users.noreply.github.com>
Signed-off-by: npub1gjuws2a2dc8z2nszprtg7v6u9q7ffeah3hgl5yx45jwn7y7aqs6s5e9xj6 <44b8e82baa6e0e254e0208d68f335c283c94e7b78dd1fa10d5a49d3f13dd0435@buzz.block.builderlab.xyz>
Co-authored-by: npub1yxv5wk0u0fh6dwt925wntn7h397jvteyj4r87ttcd9xae7n2t3lqqj9jmm <21994759fc7a6fa6b965551d35cfd7897d262f2495467f2d78694ddcfa6a5c7e@buzz.block.builderlab.xyz>
Co-authored-by: npub1gjuws2a2dc8z2nszprtg7v6u9q7ffeah3hgl5yx45jwn7y7aqs6s5e9xj6 <44b8e82baa6e0e254e0208d68f335c283c94e7b78dd1fa10d5a49d3f13dd0435@buzz.block.builderlab.xyz>
Relands **#2467** (extract `buzz-voice` crate) and **#3208** (Pocket
voice settings) onto main, after #3266 and #3180 merged.
## Why a fresh PR
The repo is squash-only with delete-branch-on-merge. Squashing #3266
deleted `jtennant/pocket-tts-2026-04`, which was #2467's base — GitHub
auto-closed #2467 and it cannot be reopened. Squash merges also sever
ancestry, so GitHub's natural merge-base reports phantom conflicts for
the whole remaining stack.
## Content provenance
- Byte-identical to the blessed `jt/buzz-voice-refactor` branch
(`93029c577`, tree `6729e0eff` — reviewed by Dawn (#2467) and Max
(#3208) at exact heads) **except** the three files where #3180 and #3208
genuinely interact.
- Three-file resolution (union of both sides):
- `huddle/mod.rs` — #3180's pipeline re-exports + #3208's
`agent_tts_routing` imports.
- `huddle/state.rs` — `reset_preserving_generation` preserves both
`huddle_generation` (#3180) and `tts_enabled` (#3208); test sets merged
into one `tests` module.
- `desktop/src/testing/e2eBridge.ts` — both switch arms kept; no
duplicate case labels.
## Verification at cf32dacd7
- `cargo test` (desktop/src-tauri, pinned 1.95.0): **2047 + 3 pass / 0
fail** (14 ignored: 8 keychain, 4 real_relay, 2 flag-gated)
- `cargo clippy --all-targets -- -D warnings`: clean; `cargo fmt
--check`: clean
- `cargo check --workspace` (root, includes new `buzz-voice` member):
clean; `cargo test -p buzz-voice`: 5/0
- `pnpm test`: **3885 / 0**; `tsc --noEmit`: clean; lint: clean
The 3180×3208 interaction resolution is getting an independent team
re-review before merge.
Buzz channel: buzz-desktop-voice `fd5fb402-b651-4238-89b1-bb3e2fa4dc96`,
thread `b4798ecc`.
Signed-off-by: npub1qyvc0c5kl4gqv2fd97fsk46tu378sqgy35vc83rvgfwne90sel7s0ed67d <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz>
Co-authored-by: npub1qyvc0c5kl4gqv2fd97fsk46tu378sqgy35vc83rvgfwne90sel7s0ed67d <011987e296fd5006292d2f930b574be47c7801048d1983c46c425d3c95f0cffd@buzz.block.builderlab.xyz>
## Summary
- report the relay as connected immediately after socket open and
successful AUTH
- keep rate-limited subscription replay, the connect promise, and
reconnect listeners unchanged
- cover authenticated reconnect while replay is held behind the shared
rate-limit gate
## Why
After WARP recovery, the socket could reopen and authenticate
successfully while subscription replay waited behind the existing
rate-limit gate. `connect()` kept `ConnectionState` at `reconnecting`
during that intentional delay, so the desktop displayed “Can’t reach the
relay” despite authenticated traffic already flowing.
This is separate from #3774: that fix keeps routine operations from
bypassing scheduled reconnect backoff. This patch preserves those
protections and only corrects the authenticated transport-state
boundary.
## Failure semantics
If replay fails after the early `connected` transition, the existing
`replayLiveSubscriptions()` catch calls `resetConnection()`, closes the
socket, returns state to `reconnecting`, and schedules recovery.
Operation waiters and reconnect notifications still do not complete
until replay succeeds.
## Validation
At commit `c8a4308e1079f4f9e6a72f0f0bfba280fe822ec0` with a clean
working tree:
- `pnpm --dir desktop typecheck`
- `pnpm --dir desktop test` — 3,847 passed
- `pnpm --dir desktop check` — passed; two pre-existing informational
template-literal notices
- `pnpm --dir desktop exec playwright test
tests/e2e/relay-reconnect.spec.ts` — 8 passed
- regression test proven red before the production ordering change
(`reconnecting` after 3 seconds) and green after it
Signed-off-by: Wes <wesbillman@users.noreply.github.com>
Co-authored-by: Carl <c7ebe626f000404285d3686e1dc74cc07cc60a9754a150041ba132e14bd3e2ec@buzz.block.builderlab.xyz>
## Summary
- make the relay reconnect coordinator authoritative during outages so
query, publish, and subscription traffic waits for the scheduled attempt
instead of cancelling backoff
- release waiting operations after the coordinated AUTH +
live-subscription replay attempt, while preserving one explicit manual
reconnect fast path
- suppress duplicate notification side effects when reconnect replay
overlaps previously delivered events
## Root cause
`resetConnection()` scheduled exponential backoff, but
`ensureConnected()` cleared any pending reconnect timer. Operation-level
retry paths immediately called `ensureConnected()`, so ordinary app
traffic could repeatedly bypass the reconnect policy during an outage.
The resulting churn also replayed overlapping live events into
notification side effects without a shared event-ID guard.
## Validation
- `pnpm --dir desktop typecheck`
- `pnpm --dir desktop test` — 3,823 passed
- pre-push: `desktop-check`, `desktop-test`, and `branch-skew` passed
- file-size, px-text, and pubkey-truncation ratchets passed
---------
Signed-off-by: Wes <wesbillman@users.noreply.github.com>
Co-authored-by: Carl <c7ebe626f000404285d3686e1dc74cc07cc60a9754a150041ba132e14bd3e2ec@buzz.block.builderlab.xyz>
## Summary
- align the multi-community rail with the content surface and balance
its visible 10px side gutters
- center Mobile pairing, start sessions on demand, and keep retry states
inside the QR area
- reveal the QR code and copy action with 250ms motion and use the
standard loading spinner
The rail was centered within its own box, but the adjacent sidebar added
another 11px to the visible right gap. Mobile pairing also started
before user intent, which could leave an idle session waiting for EOSE.
## Validation
- `pnpm -C desktop build:e2e`
- `pnpm -C desktop test` — 3,516 passed
- `pnpm -C desktop exec playwright test tests/e2e/community-rail.spec.ts
--project=smoke` — 19 passed
- `pnpm -C desktop exec playwright test
tests/e2e/mobile-pairing-qr.spec.ts --project=smoke` — 1 passed
`pnpm -C desktop check` is currently blocked by the existing
`src-tauri/src/managed_agents/runtime.rs` file-size baseline (2,220
lines; limit 2,216).
