id: poc: bridge sync session — native store ↔ DO-shaped SQLite store over WebSocket, fork-proof exchange + withholding detection (from #2058) #2276

Activity

1. dezren39

   added

   app/idPackage/app: id (pkgs/ or apps/ in this repo)Package/app: id (pkgs/ or apps/ in this repo)

   on Oct 29, 2026
2. developing-today commented on Oct 29, 2026

   

   developing-todayBot

   on Oct 29, 2026 – with developing.today

   Contributor

   More actions

   CODE-1137
3. dezren39

   added

   queue/queuedHas a queued task in the Supabase agent queueHas a queued task in the Supabase agent queue

   on Oct 29, 2026
4. github-actions

   assigned

   dezren39

   on Oct 29, 2026
5. dezren39

   added

   kind/enhancement

   source/humanFiled directly by a humanFiled directly by a human

   priority/normalOrdinary priority (the default)Ordinary priority (the default)

   effort/mMedium: expected under 1 day of workMedium: expected under 1 day of work

   status/needs-triageNot yet triaged; needs a first lookNot yet triaged; needs a first look

   meta/enrichedIntake review done: acceptance criteria, links, and normalized labels addedIntake review done: acceptance criteria, links, and normalized labels added

   on Oct 29, 2026
6. dezren39 commented on Oct 29, 2026

   

   dezren39

   on Oct 29, 2026

   MemberAuthor

   More actions

   Intake review — fleet monitor, 2026-10-29

   • Related: bridge-semantics investigation on id: investigate: bridge semantics — hosted (HTTPS/WS) edition syncing with native Iroh nodes #2058; epic epic: ID platform expansion — apps, comms, currency, hosting, boards (2026-10-28 directive, LIST 6–15) #2056. Already carries queue/queued — a live queue task exists for it (verified in this run's reconciliation).
   • Still live: filed today as a PoC; no bridge sync session exists yet.
   • Notes: body already carries acceptance criteria; none appended. Fork-proof exchange is the property under test — the PoC should state explicitly which fork case it exercises.
   • Risks: medium — WebSocket session semantics over two store shapes can sprawl; the issue's own "smallest thing that exercises the real rules" scoping is the guard.
   • Suggested approach: reuse the native store machinery unchanged (per the issue), one session, one fork case, then report back to id: investigate: bridge semantics — hosted (HTTPS/WS) edition syncing with native Iroh nodes #2058.
   • Labels normalized: kind/enhancement, source/human, priority/normal, effort/m, status/needs-triage, meta/enriched.
7. dezren39

   added

   queue/claimedQueue worker lifecycle stateQueue worker lifecycle state

   work/assignedQueue worker lifecycle stateQueue worker lifecycle state

   and removed

   status/needs-triageNot yet triaged; needs a first lookNot yet triaged; needs a first look

   queue/queuedHas a queued task in the Supabase agent queueHas a queued task in the Supabase agent queue

   on Oct 30, 2026
8. 32 remaining items

   Load more
9. dezren39

   added a commit that references this issue on Oct 30, 2026

   test(id): bridge sync session PoC — native store ↔ DO-shaped SQLite s…

   ...

   b3421ef
10. dezren39 commented on Oct 30, 2026

    

    dezren39

    on Oct 30, 2026

    MemberAuthor

    More actions

    🤖 Agent note · done (PoC delivered; PR awaiting CI)

    • Agent: queue worker (repo-worker-code), lane C shift
    • Session / ID: queue task 3431
    • Found via: queue claim
    • Task: Bridge sync session PoC per this issue's acceptance — delivered in PR test(id): bridge sync session PoC — native store ↔ DO-shaped SQLite store over WebSocket (#2276) #2848 (pkgs/id/tests/bridge_poc.rs, test-only)
    • Links: PR test(id): bridge sync session PoC — native store ↔ DO-shaped SQLite store over WebSocket (#2276) #2848 · branch queue/3431-2276-bridge-sync-poc · commit b3421ef

    Findings for the #2046 Q1 decision (fork-proof cadence)

    Measured on the PoC (laptop run, debug build; all 5 PoC tests pass, 10/10 with the shared test-common unit tests):

    • Fork set (2 authors' worth of state: 3 events, one fork at equal heights): proof-carrying range exchange = 4 flights, 163 reconciliation bytes — and it detected the fork that a height comparison provably misses (pre-session chains had identical next_seq, different heads).
    • Bootstrap set (100 events / 4 authors, one side empty): 4 flights, 3,235 reconciliation bytes ≈ 32.4 bytes/event — about half the naive bound of mailing every event ID both ways (6,400 B), and the flight count did not grow between the 3-event and 100-event sets at this fan-out. Session wall time 5.3 s in a debug build is dominated by per-event signature verification in ingest_batch, not by the exchange.
    • Cadence read: at wave-1 scale the exchange itself is cheap (bytes scale ~linearly at ~32 B/event with fingerprint flights, flights stayed at 4); the cost a cadence decision actually budgets is the verification of transferred events, which is paid only for the symmetric difference. Nothing in the measurements argues against running the proof exchange on every bridge session at this scale; the open part of Q1 remains the DO rows/CPU accounting on the Workers side, which this PoC does not measure (no Workers runtime here).

    What the PoC proved, per acceptance

    • Convergence: push+pull in one session → identical applied sets (store canonical state) and identical live tag claims on both stores.
    • Forks: a fork authored on each side while disconnected is detected by the proof-carrying exchange (≥1 Ingest::Fork outcome per side), and both sides end holding the identical fork proof (same seq, same first/second IDs).
    • Withholding: a relay that drops one author's events is detected from the announced chain head (seq 4 / head hash) plus the after-frontier reference inside a signed event that did arrive (the receiver holds that event waiting on exactly the withheld hash). Detection names the author, the announced head, and the undelivered event IDs; the honest-session control produces no evidence, and an author whose events arrived but waits on the withheld dependency is not itself flagged.
    • Replay identity: the same journal replayed into a fresh native store and a fresh hosted (SQLite) store — in opposite arrival orders — produces identical tables (events / chains / forks / proof_ids / proof_headers compared row-for-row).

    Limitations (stated in the PR as well)

    • The sync set is enumerated via the store's public read API, which exposes the lowest fork proof per author; the internal sync set also carries higher fork points' proof headers.
    • The hello's chain heads are unsigned in the PoC; production evidence for the head half of withholding detection arrives inside signed events (backlinks / after), which is the half the frontier test exercises with real signatures.
    • The hosted store's SQLite tables are mirrored from the same ingest outcomes in the same step (one batch = one transaction, the DO write shape); a rusqlite-native ingest port is separate work, not needed for the semantics questions this PoC answers.
11. dezren39 commented on Oct 30, 2026

    

    dezren39

    on Oct 30, 2026

    MemberAuthor

    More actions

    🤖 Agent note · parked

    • Agent: queue worker (repo-worker-code), lane C shift
    • Session / ID: queue task 3431
    • Found via: queue claim
    • Task: PoC delivered in PR test(id): bridge sync session PoC — native store ↔ DO-shaped SQLite store over WebSocket (#2276) #2848 and findings posted on this issue; parked pending CI on the PR (resume: read PR checks fresh, disposition any Codex findings, merge under the standing criteria, then complete task 3431).
    • Links: PR test(id): bridge sync session PoC — native store ↔ DO-shaped SQLite store over WebSocket (#2276) #2848
12. dezren39

    added

    queue/parkedQueue task is parked (waiting on CI, time, or an event)Queue task is parked (waiting on CI, time, or an event)

    status/parkedDeliberately paused with a recorded resume conditionDeliberately paused with a recorded resume condition

    and removed

    status/in-progressQueue worker lifecycle stateQueue worker lifecycle state

    queue/claimedQueue worker lifecycle stateQueue worker lifecycle state

    on Oct 30, 2026
13. dezren39

    added 2 commits that reference this issue on Oct 30, 2026

    test(id): address Codex review on bridge PoC — waiting-event snapshot…

    ...

    d14d26b

    Merge pull request #2848 from developing-today/queue/3431-2276-bridge…

    ...

    d2eb34e
14. dezren39 commented on Oct 30, 2026

    

    dezren39

    on Oct 30, 2026

    MemberAuthor

    More actions

    🤖 Agent note · done

    • Agent: queue worker (repo-worker-code), openchamber scheduled lane
    • Session / ID: queue task 3431
    • Task: Bridge sync session PoC per this issue acceptance (id: poc: bridge sync session — native store ↔ DO-shaped SQLite store over WebSocket, fork-proof exchange + withholding detection (from #2058) #2276) — delivered and merged in PR test(id): bridge sync session PoC — native store ↔ DO-shaped SQLite store over WebSocket (#2276) #2848 (merge commit d2eb34ed9d344a8cebd8e7caf0e9071ba38fbbae).
    • Codex review resolution:
      1. Added EventStore::waiting_event to look up and decode events directly from WAITING rows, updated held_events to retrieve waiting events, and added bridge_session_converges_with_waiting_events_on_both_sides.
      2. Added sparse_fork_reconciliation_scales_sublinearly_across_set_sizes demonstrating that for a single-fork difference against shared histories of 20, 60, and 100 events, exchange flights remain constant (5) and bytes grow sublinearly (1,036 to 1,356 bytes).
    • Verification: 12/12 tests in bridge_poc.rs passed cleanly on laptop; PR merged into main.
    • Links: PR test(id): bridge sync session PoC — native store ↔ DO-shaped SQLite store over WebSocket (#2276) #2848 · merge commit d2eb34ed9d344a8cebd8e7caf0e9071ba38fbbae
15. dezren39

    closed this as completedon Oct 30, 2026
16. dezren39

    removed

    status/parkedDeliberately paused with a recorded resume conditionDeliberately paused with a recorded resume condition

    queue/parkedQueue task is parked (waiting on CI, time, or an event)Queue task is parked (waiting on CI, time, or an event)

    work/assignedQueue worker lifecycle stateQueue worker lifecycle state

    on Oct 30, 2026
17. dezren39

    added a commit that references this issue on Oct 31, 2026

    fix(id): make the no-SQLite guard read the resolved graph, not the ra…

    ...

    6aaef4b