Skip to content

refactor(dashboard): route chat recovery replays through one ledger - #17067

Merged
chenmingwei23 merged 1 commit into
mainfrom
refactor/w2-02-recovery-replays
Oct 6, 2026
Merged

chenmingwei23 merged 1 commit into
mainfrom
refactor/w2-02-recovery-replays

Conversation

@bolichen97

@bolichen97 bolichen97 commented Oct 5, 2026 •

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: A dashboard chat's self-queued recovery replays are recorded, claimed and cancelled through one ledger on the slot, so a Stop, a rebind or a newer message cancels every replay family by the same rule.

The dashboard re-queues a turn for itself after five kinds of recoverable failure: the model-access swap, the lost-session reconnect, the image-history recovery, the content-filter retry and the auto-continuations. Each kind kept its own three to five private attributes on _ChatSlot, and the "drop it if the user stopped, moved or replaced it" rule was written out again for each kind at the queue drain, the dispatch match, the consume seam and eight arm sites. The lost-session copy had already drifted into its own re-implementation of the shared rule.

Why it matters

A replay that slips past this rule re-runs a request the user cancelled: it can repeat a destructive tool call or answer into the wrong session. With the rule spread across about twenty sites in a file that changes daily, a new replay kind needed six coordinated edits, and missing one clause (the in-flight Stop, the session-scoped count, the rebind) failed silently. One module with one rule and one family registry makes that clause impossible to forget: a new replay kind becomes a ReplayFamily member, a POLICIES row (its position is the drain's re-check order, its consume_phase the consume-seam slot) and the arm site that queues it.

Not a goal

  • No behaviour change beyond two ruled log-only differences: the lost-session drain step drops a duplicate "Suppressing re-queue" log line, and cancel log lines print ReplayRevocation(...) instead of _ReplayRevocation(...). Every family keeps its notices, refunds, ordering and timing.
  • The one-shot allowances (_model_access_fallback_used, _poisoned_reset_used, _session_not_found_retry_used, _refusal_fallback_attempted, _promise_only_retries, _compaction_continue_retries) stay on the slot; the ledger only names which one a cancellation re-arms.
  • The refusal swap's binding (_refusal_fallback_session_key) stays owned by chat_turn/model_fallback.py; the content-filter record copies it at arm.
  • No further split of _run_chat or _start_next_queued_turn beyond their replay sites.

What changed (motivation → approach → change)

Every replay family had the same shape (an entry id, a session binding, two Stop counts) and the same question to answer twice, so the shape and the question move into one deep module, dashboard/recovery_replays.py, owned by slot.replays. POLICIES is the one registry of families: whether one queue entry identifies a family, where the consume seam checks it, its notice wording, which one-shot a cancellation refunds and whether an accepted replay keeps its record are fields of its row, and ENTRY_FAMILIES is derived from it in declared order. The continuation notice's trigger-priority wording stays in cancel_notice.

Seam Before After
State on _ChatSlot 19 replay attributes in 5 families one replays: RecoveryReplays
Arm sites (8 in chat_runner.py) 3-5 attribute writes each slot.replays.arm(family, ...)
Drain (_start_next_queued_turn) an entry-gone guard, 3 drop functions and an inline lost-session block _forget_swept_replays + _drop_revoked_replays
Dispatch match 4 copy-pasted drained-alone blocks slot.replays.claim(consumed)
_run_chat 4 replay booleans one _replay: frozenset[ReplayFamily]
Consume seam 3 veto functions and an inline lost-session block _replay_vetoed_at_consume(state, slot, claimed, phase), phase read from the policy
Continuations _purge_superseded_continuations with its own snapshot reads the same purge over the ledger's continuation record

The ledger never reads the slot itself: a re-check takes a LiveSlot (binding, Stop counts, Stop in flight, waiting user input) that chat_turn/recovery.py reads, so the rule is a pure function. The appliers stay in that composed owner because _purge_superseded_continuations must keep the chat_runner._should_suppress_requeue patch target working; the other two appliers need no patched runner name and moving them out is a logged follow-up. The owner imports the ledger from its own module, and chat_runner re-exports one ledger name for live importers (SESSION_NOT_FOUND_CANCELLED_TEXT, which tests import from the runner).

Interface (slot.replays: RecoveryReplays):

Entry point Behaviour, invariants, errors
arm(family, *, entry_id, session_key, stop_gen, session_stop_gen) replaces the family's record; an empty entry_id disarms an entry family; never raises
claim(consumed) -> frozenset[ReplayFamily] the families whose entry drained alone; a merged dispatch claims nothing; one entry can carry two families
revalidate(family, live) -> ReplayRevocation pure; rebound / Stop moved / Stop in flight / user input; an empty recorded binding is never a rebind; an unarmed family reads as not revoked
sweep(live_ids, families) -> [(family, entry_id)] forgets records whose entry left the queue and returns them, so the caller re-arms that family's sweep_refunds
armed / entry_id / disarm read or forget one record
replays_of(slot) the slot's ledger, attaching an empty one to a partially built slot; a slot that cannot hold one raises AttributeError (fail closed)
POLICIES / ENTRY_FAMILIES / consumed_in(phase) the registry, the entry families in drain order, the families a consume phase checks

Configuration: none. Ordering: the drain re-checks families in registry order (ENTRY_FAMILIES, pinned by a test); the consume seam checks the lost-session and image families before the turn-start allowance refresh and the model-access and content-filter families after it, as each row's consume_phase says.

Deletion test: delete recovery_replays.py and the 19 attributes return to _ChatSlot, the revocation rule returns as two implementations plus three inline copies, the drained-alone match returns four times in the drain, the per-family notices and refunds return to eight functions in chat_turn/recovery.py, and _run_chat regains four parameters: five families times the drain, dispatch and consume seams, plus eight arm sites.

Intentional log-only differences (ruled): the lost-session drain step no longer logs a second "Suppressing re-queue" line right after the continuation purge logged the same line; the revocation repr inside cancel log lines reads ReplayRevocation(...) instead of _ReplayRevocation(...).

Docs: learn-cron-dashboard.md lists the ledger among the slot's components, notes it as the one slot-owned container behind its own interface, and its recovery row gives the recipe for a new family; of the three citations of deleted names in error-handling.md and model-fallback.md (coordinator grant), two now name RecoveryReplays.revalidate and one names _run_chat's _replay.

Backwards compatibility

Compatible: every removed name is private to the dashboard turn runtime, and its only producers and readers are migrated in this diff. SESSION_NOT_FOUND_CANCELLED_TEXT still imports from kiro_crew.dashboard.chat_runner.

Reader: src/kiro_crew/dashboard/chat_runner.py:_start_next_queued_turn -- chat -- the only producer of _run_chat's replay flags now passes _replay (test_recovery_replays.py::test_the_drain_claims_a_replay_that_may_run)
Reader: src/kiro_crew/dashboard/chat_runner.py:_run_chat -- chat -- the consume seam reads _replay (test_recovery_replays.py::test_the_consume_seam_vetoes_a_stopped_replay_before_the_provider)
Reader: src/kiro_crew/dashboard/chat_turn/model_fallback.py:_restore_refusal_fallback -- chat -- reads _refusal_fallback_session_key, which stays on the slot (test_dashboard_chat.py::TestRunChatRefusalFallback::test_restore_dropped_when_slot_rebound)
Reader: src/kiro_crew/dashboard/chat_regenerate.py:api_chat_slot_edit_resend -- chat -- its shallow slot copy shares replays but only appends rows (test_chat_regenerate_cov80.py)
Reader: src/kiro_crew/dashboard/chat_rewind.py:api_chat_slot_rewind -- chat -- same shallow copy, same rows-only use (test_dashboard_chat_rewind.py)

Tests

Test Locks in
test/test_recovery_replays.py (new, 89 tests) the registry (every family has one POLICIES row, entry families in pinned drain order, each in exactly one consume phase, every refund names a real _ChatSlot slot); one row per family: arm then claim alone vs merged, two families on one entry, Stop moved / Stop in flight / rebind / user input, the recorded vs live session count, sweep and its refund, notices, replays_of failing closed; and both seams through a real slot (drain drop + refund, drain claim, consume veto + refund, accepted record kept only for the content filter)
owned suites migrated test_dashboard_chat.py, test_model_access_fallback_interactive.py, test_chat_session_not_found_recovery.py, test_refusal_replay_consume_abort_chat_done.py, test_promise_only_recovery.py, test_cross_surface_stop_gate.py: private attribute writes become slot.replays.arm(...), reads become armed() / entry_id() / revalidate() assertions; TestRunChatRefusalFallback gains an end-to-end refuse -> swap -> enqueue -> relink -> drain test and a rebind check on the requeue carry, pinning both content-filter arm sites
test_chat_runner_composition_contract.py base-name and surface rows of the deleted helpers removed; the two appliers join _AWAIT_FREE_PHASES; the _run_chat signature pin names _replay

Replaced, not layered: the two direct unit tests of _refusal_replay_vetoed_at_consume now drive _replay_vetoed_at_consume, the seam every family shares. Mutation proofs: 22 mutations, each turning a named test red. They break the rebind clause, the empty-binding rule, the drain-head sweep, the drained-alone claim, each family's session-count key, the content-filter record's lifetime, the model-access refund, the in-flight Stop, user input, the session-scoped count, the consume seam, a continuation arm, the two-family claim, the await-free drain, the binding at both content-filter arm sites, the registry's keyed flag, consume phase and order, a refund naming a missing slot, and replays_of failing closed.

Evidence, recorded on this exact base:

  • BASE-vs-HEAD differential on base c6bfd72, 466 drain / consume / continuation scenarios through the real seams: identical except the 8 ruled duplicate-log cases (the logs field only).
  • Owned + adjacent manifest (80 files, rebased head): 7,689 passed, 1 skipped. An earlier per-node run against BASE on 77 shared files differed only by the replaced pin rows and by environment errors that happen only in the BASE tree.
  • scripts/local-gate.py --dry-run selects 1,973 backend files, so per the campaign capacity rule the related set runs in CI.

Manual verification

N/A — unit coverage sufficient: the change is internal to the turn runtime and both seams are driven through real _ChatSlots in the tests and the differential.

Related Issues

Part of #16536

Checklist

  • At most two commits (one is the norm), with a Conventional Commits title (feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)
  • Existing tests pass and new tests added for new functionality
  • Self-review completed; code follows project style guidelines
  • Documentation updated (if applicable)
  • No secrets, credentials, or internal references in the diff

@bolichen97
bolichen97 requested a review from a team as a code owner October 5, 2026 16:45
@github-actions github-actions Bot added the readiness: checking Automated validation is still running label Oct 5, 2026
@github-actions

github-actions Bot commented Oct 5, 2026 •

Copy link
Copy Markdown
Contributor

Design Review (Opus 5.5) — ✅ PASS

Design-level review of e38028f59a80d1530f69fa42109d2011b2f28ab4 — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

Design-Verdict: PASS

Check that _drop_revoked_replays covers the _should_suppress_requeue(slot) clause the old lost-session drain dropped on, via the earlier continuation purge or revalidate.

[DESIGN-REVIEWED] e38028f

Description read

[DESCRIPTION-READ] 2813cf38da6ee436b2ec4d66a27aa8a790da02e7f945d56a090216a88c5fea74

That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with .github/scripts/pr-description-capture.sh over the same inputs; if it differs, the description or that evidence changed after this verdict was formed and any finding drawn from either is unproven.

@github-actions

github-actions Bot commented Oct 5, 2026 •

Copy link
Copy Markdown
Contributor

First Principles Review (Opus 5.5) — 🟡 CONCERNS

Premise-level review of e38028f59a80d1530f69fa42109d2011b2f28ab4 — why this exists and whether the shipped surface is the smallest honest version. Updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

First-Principles-Verdict: CONCERNS

ReplayPolicy.log_flags is a knob with one consumer. It exists only to keep one family's log-line format, and the Goal already accepts log-only differences.

Not justified as shipped

  • Item 5 — one consumer, generalized: log_flags: bool = False is set by exactly 1 of 5 POLICIES rows (SESSION_NOT_FOUND, log_flags=True). It keeps a (stop_moved, superseded, rebound) log format. The description already rules log-only drift acceptable: "No behaviour change beyond two ruled log-only differences". With the knob gone, the lost-session cancel log prints ReplayRevocation(...) like the other three families, and no user or code path notices. Grep log_flags: 1 reader (log_args), 1 setter.
    Subtraction: delete ReplayPolicy.log_flags and log_args. Make both lost-session log strings single-%s, and pass revocation directly at the drain and consume sites.
    Clears when: the field and log_args are removed, or a log consumer that parses the three-flag format is named.

What this change ships

Inventory (6 items) — 5 justified

Intent: one rule decides when a Stop, a rebind or a newer message cancels every self-queued recovery replay. This is a FIX-shaped refactor.

  1. 19 replay attributes on _ChatSlot replaced by one slot.replays ledger — justified
  2. One revocation rule (revalidate) replaces several copies, including the lost-session copy that had drifted — justified
  3. One drained-alone claim replaces four copy-pasted matches in the dispatch path — justified
  4. A per-family POLICIES registry (consume phase, refunds, sweep point, keep-record) keeps each family's existing timing — justified
  5. A per-family log-format switch (log_flags) — one consumer, generalized: preserves a log shape the Goal already lets drift
  6. A duplicate "Suppressing re-queue" log line is dropped and the repr is renamed — justified

[FIRST-PRINCIPLES-REVIEWED] e38028f

Description read

[DESCRIPTION-READ] 7e4ac605c20cfd85c0d1a91532c322b8821c73352f240a774544f502daf35bb5

That sha256 names the PR description this verdict read. Recompute it with .github/scripts/pr-description-capture.sh; if it differs, the description changed after this verdict was formed and any finding drawn from the description is unproven.

@github-actions

github-actions Bot commented Oct 5, 2026 •

Copy link
Copy Markdown
Contributor

Opus 5.5 Review — ✅ no blocking findings

Reviewed e38028f59a80d1530f69fa42109d2011b2f28ab4 — this comment is updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] e38028f

Verdict parsed from the review's SHA-scoped output markers for commit e38028f59a80d1530f69fa42109d2011b2f28ab4.

False positive or not applicable? A repository writer can comment:
/ai-review override fable e38028f59a80d1530f69fa42109d2011b2f28ab4: <one-sentence reason>

@github-actions

github-actions Bot commented Oct 5, 2026 •

Copy link
Copy Markdown
Contributor

GPT 6.1 Review — ✅ no blocking findings

GPT 6.1 completed its review of e38028f59a80d1530f69fa42109d2011b2f28ab4 and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] e38028f

False positive or not applicable? A repository writer can comment:
/ai-review override gpt e38028f59a80d1530f69fa42109d2011b2f28ab4: <one-sentence reason>

@github-actions github-actions Bot added readiness: passed Eligible automated validation passed for the current revision and removed readiness: checking Automated validation is still running labels Oct 5, 2026
@bolichen97
bolichen97 force-pushed the refactor/w2-02-recovery-replays branch from 17a77db to 942c06e Compare October 5, 2026 19:38
@bolichen97

Copy link
Copy Markdown
Contributor Author
  • span=dc2fd0d36732 — fixed: the zero-consumer text field is deleted

fixed in 942c06e: the record's text field, arm(..., text=) and RecoveryReplays.text() are gone; the content-filter arm passes only the entry id, binding and Stop counts.
BASE's _refusal_retry_text was already write-only: written at one arm site and cleared at the others, never read in production, so dropping it changes no observable behaviour (the BASE-vs-HEAD differential over 466 drain / consume / continuation scenarios is unchanged).

@bolichen97

Copy link
Copy Markdown
Contributor Author
  • Watch: every refund attribute in POLICIES must be a real _ChatSlot slot

fixed in 942c06e: test_recovery_replays.py::test_every_refund_names_a_real_slot_allowance asserts every drain / consume / sweep refund name is in _ChatSlot.__slots__, so a renamed one-shot fails on every run instead of only when a replay is cancelled.
Mutation proof: renaming _poisoned_reset_used in the image-history row turns that test red. _ChatSlot has no __dict__, so a stale name also raises AttributeError at the cancellation itself rather than writing a stray attribute.

@github-actions github-actions Bot added readiness: checking Automated validation is still running and removed readiness: passed Eligible automated validation passed for the current revision labels Oct 5, 2026
@bolichen97

Copy link
Copy Markdown
Contributor Author
  • Watch: dropping _refusal_retry_text must keep the content-filter replay's identity at dispatch, merged dispatches included

rebutted (verified, no change needed): on BASE the dispatch already matched the content-filter replay by queue-entry id, not text. chat_runner.py@c6bfd72db L7109-7115 compares consumed[0].get("id") with _refusal_replay_queue_id only when len(consumed) == 1. _refusal_retry_text was written at one arm site and cleared at six, never read; the state.py comment saying "the runner matches it at dispatch" was stale.
HEAD keeps that rule in RecoveryReplays.claim: a merge returns no family. test_recovery_replays.py::test_an_armed_replay_is_claimed_only_when_its_entry_drained_alone[CONTENT_FILTER] pins alone vs merged. The 466-scenario BASE-vs-HEAD differential, which records the claimed families per drain with and without a second queued entry, is identical apart from the 8 ruled log lines.

@github-actions github-actions Bot added readiness: action required A blocking check or review needs attention and removed readiness: checking Automated validation is still running labels Oct 5, 2026
@bolichen97
bolichen97 force-pushed the refactor/w2-02-recovery-replays branch from 942c06e to 7cfa31f Compare October 6, 2026 00:25
@github-actions github-actions Bot added readiness: checking Automated validation is still running and removed readiness: action required A blocking check or review needs attention labels Oct 6, 2026
@bolichen97

Copy link
Copy Markdown
Contributor Author
  • span=97f11491b41a — rebutted: log_flags keeps the lost-session log lines byte-identical to BASE

log_flags / log_args reproduce BASE's two lost-session cancel lines exactly: chat_runner.py@c6bfd72db L6647 (stop_since_enqueue=%s superseded=%s rebound=%s) and the consume line at L8913. The campaign's behaviour-preservation rule (W5) bars an observable change, log wording included, without a recorded ruling. The two log-only differences in the description are the ruled ones; this wording is not among them.
Dropping the knob is a log-format change, not a subtraction of dead code, so it is out of scope for this refactor. Proposed follow-up: unify every cancel line on the ReplayRevocation repr and delete log_flags/log_args in a PR whose goal is that wording change.

@bolichen97

Copy link
Copy Markdown
Contributor Author
  • Watch: the old lost-session check must equal the shared revalidate rule, empty bindings and the Stop-count key included

rebutted (verified): BASE _session_not_found_replay_revoked (chat_runner.py@c6bfd72db L6293-6313) agrees with revalidate clause by clause. Rebound is bool(bound_key) and live_key != bound_key, so an empty binding is never a rebind. The session Stop count is read for bound_key or live_key, which is session_count_of="recorded": record.session_key or live.session_key. Stopped means the slot or session counter moved; superseded means a pending steer or a user follow-up.
Its Stop-in-flight clause lived in the caller (_should_suppress_requeue, i.e. _stop_state != "idle") and is now stopping; that split was the drift. The BASE-vs-HEAD differential runs this family through drain and consume under every trigger with both a recorded and an empty binding: identical apart from the ruled duplicate log line.

@github-actions github-actions Bot added readiness: action required A blocking check or review needs attention and removed readiness: checking Automated validation is still running labels Oct 6, 2026
@bolichen97

Copy link
Copy Markdown
Contributor Author
  • span=2f0f311215c7 — rebutted: the replays_of attach branch has live readers outside this PR

Measured, not assumed: with the branch deleted (return slot.replays), the 80-file owned + adjacent manifest loses 8 tests in test/test_prompts.py, a file this PR does not own: TestRunChatPrompts (7) and TestPromptExpansionStaysOffTheEventLoop (1). Each fails AttributeError: '_Slot' object has no attribute 'replays' from _run_chat -> _replay_vetoed_at_consume (chat_turn/recovery.py L488). Their _Slot stub (L217) predates the ledger, and the campaign keeps unowned tests that drive owned code read-only.
BASE gave those callers the same tolerance through per-attribute getattr(slot, "_x", default) reads at each replay site. The branch is that tolerance in one place, and a slot that cannot hold a ledger still fails closed (test_replays_of_refuses_a_slot_that_cannot_keep_its_ledger). Deleting it means rewriting that unowned stub, which is a follow-up for its owner.

@bolichen97

Copy link
Copy Markdown
Contributor Author

Re-ran job 112045813315 (Backend Tests (Windows) (2)) once. Its only failure, test/test_dashboard_chat.py::TestFolderCRUD::test_resume_session_unhides_folder (409 resume_conflict), is the recorded Windows clock-tie flake #17117; #17145 is the fix in flight. This PR does not touch that test or the resume path.

The dashboard re-queues a turn for itself after five recoverable
failures (model-access swap, lost-session reconnect, image-history
recovery, content-filter retry, auto-continuations). Each family kept
three to five private attributes on _ChatSlot and its own copy of the
"cancel on Stop, rebind or newer message" rule at the queue drain, the
dispatch match, the consume seam and eight arm sites.

dashboard/recovery_replays.py is the new deep module behind
slot.replays: arm, claim (the drained-alone rule; one entry can carry
two families), sweep, a pure revalidate over a LiveSlot view, and one
policy table for the per-family notices, refunds and record lifetime.
chat_turn/recovery.py keeps the two seam appliers, so every chat_runner
patch target still reaches them. _run_chat takes one _replay frozenset
in place of four booleans.

Behaviour is unchanged apart from two ruled log-only differences: the
lost-session drain step no longer logs a duplicate "Suppressing
re-queue" line, and cancel log lines print ReplayRevocation(...).

Part of #16536
@bolichen97
bolichen97 force-pushed the refactor/w2-02-recovery-replays branch from 7cfa31f to e38028f Compare October 6, 2026 02:21
@github-actions github-actions Bot added readiness: checking Automated validation is still running and removed readiness: action required A blocking check or review needs attention labels Oct 6, 2026
@bolichen97

Copy link
Copy Markdown
Contributor Author
  • span=b38cc6da2195 — rebutted: log_flags keeps the lost-session log lines byte-identical to BASE (same class as 97f11491b41a on 7cfa31f)

log_flags / log_args reproduce BASE's lost-session cancel lines exactly: chat_runner.py@c6bfd72db L6647 (stop_since_enqueue=%s superseded=%s rebound=%s) and the consume line at L8913. The two log-only differences the description accepts are the ones that were ruled; this wording is not among them, and the campaign's behaviour-preservation rule (W5) bars any other observable change, log wording included, without a ruling. The coordinator ruled this rebut-and-defer for this PR.
Deleting the knob is a log-format change, not removal of dead code. Follow-up: unify every cancel line on the ReplayRevocation repr, deleting log_flags/log_args, in a PR whose goal is that wording.

@bolichen97

Copy link
Copy Markdown
Contributor Author
  • Watch: _drop_revoked_replays must cover the _should_suppress_requeue(slot) clause the old lost-session drain dropped on

rebutted (verified): _should_suppress_requeue is slot._stop_state != "idle" (chat_turn/recovery.py L324-329). _replay_live reads stopping=slot._stopping, and _ChatSlot._stopping returns exactly self._stop_state != "idle" (state.py L4179-4180). revalidate sets stopping=live.stopping, so _drop_revoked_replays drops the lost-session replay on the same condition. Before the family steps the drain also runs the continuation purge, which still calls _should_suppress_requeue.
Pinned by test_recovery_replays.py::test_revalidate_names_what_happened_since_the_replay_was_queued[stop-in-flight-...SESSION_NOT_FOUND]. The BASE-vs-HEAD differential's drain:SESSION_NOT_FOUND:stopping:* and drain:CONTENT_FILTER+SESSION_NOT_FOUND:stopping:* scenarios produce the same queue, rows, notices and refunds. They differ only in BASE's second "Suppressing re-queue" log line: the ruled duplicate.

@github-actions github-actions Bot added readiness: passed Eligible automated validation passed for the current revision and removed readiness: checking Automated validation is still running labels Oct 6, 2026
@bolichen97
bolichen97 enabled auto-merge (squash) October 6, 2026 04:41
@iamwhatever

Copy link
Copy Markdown
Contributor

Reviewed for merge triage, not approving from my side: ~1.7k added / ~1k removed lines routing chat recovery replays through one ledger. Replay decides what history a recovered session re-sends, and at this size it needs a maintainer's own review. Auto-merge is already on, so a single approval merges it; please make sure a maintainer reviews it directly. CI is green.

@chenmingwei23
chenmingwei23 merged commit 2002653 into main Oct 6, 2026
101 checks passed
@chenmingwei23
chenmingwei23 deleted the refactor/w2-02-recovery-replays branch October 6, 2026 05:55
@github-actions github-actions Bot removed the readiness: passed Eligible automated validation passed for the current revision label Oct 6, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants