Repository navigation
id(world): F11.16 replay self-check (id world verify) #544
Description
What and why
id world verify replays a journal twice, compares records and reports divergence, as the determinism guard; exit: a deliberately nondeterministic guest fails. Determinism is assumed by replay, fixtures and future replication, but nothing checks it. A double replay is the cheapest guard.
Source
.opencode/plans/phase-11-foundations.md:59.opencode/plans/2026-10-20_phase-11-status.md:26doc/2026-10-17T09-00-00Z_feature_world_fixtures/
Current status
Partial. Record/replay fixtures exist (WorldCommand Record/Replay in pkgs/id/src/cli.rs, pkgs/id/src/world_fixture.rs 1,142 lines; PR #239 merged). There is no Verify variant in the WorldCommand enum (cli.rs:1388) and no double-replay comparison.
Acceptance criteria
-
id world verifyreplays a journal twice, compares the records and reports the first divergence - A deliberately nondeterministic guest fails verify; the checked-in examples pass
- It runs against the feat(id): record world sessions to replayable fixtures #239 fixtures
- How host-originated events (Host-originated events are not retained in the event stream #242) affect verification is documented
Size
M
Prerequisites
None.
Metadata
Metadata
Assignees
Labels
Package/app: id (pkgs/ or apps/ in this repo)Medium: expected under 1 day of workNew feature or capability requestIntake review done: acceptance criteria, links, and normalized labels addedOrg Project gate: active priority — orchestrator adds to the boardImportant and time-sensitiveFiled directly by a human
Activity
docs(id): Phase 14 plan — close-out: finish, measure, verify, open th…
dezren39 commentedon Oct 26, 2026
Intake review
Verified still live on main
14dd66b6(2026-10-26, fleet monitor).WorldCommandinpkgs/id/src/cli.rs(enum at :1398) hasRecord(:1858) andReplay(:1880) but noVerifyvariant, and there is no double-replay comparison anywhere in the crate. The #239 record/replay fixtures the body builds on exist.Related: PR #239 (fixtures, merged — cited in the body); host-originated events (#242) are the open design question for what verification must tolerate.
Risks / suggested approach: implement
VerifybesideReplayreusing the fixture harness: replay the journal twice through the same guest, compare the record streams, report the first divergence with its index. The nondeterministic-guest failure case in the acceptance criteria is the test that proves the guard works — build it first.41 remaining items
dezren39 commentedon Oct 27, 2026
🤖 Agent note · picked up
id world verifyreplay self-check (F11.16) (#544) #1468 CI now fully green (31 pass, 0 fail, 0 pending, mergeable CLEAN); merging under standing directive, then completing task.id world verifyreplay self-check (F11.16) (#544) #1468 merged, issue id(world): F11.16 replay self-check (id world verify) #544 closed.id world verifyreplay self-check (F11.16) (#544) #1468 · branch queue/1372-544-world-verifyfeat(id): add `id world verify` replay self-check (F11.16) (#544) (#1468
dezren39 commentedon Oct 27, 2026
🤖 Agent note · done
id world verifyreplay self-check (F11.16) (#544) #1468 merged (squash, merge commit 931f177) after CI went fully green (31 pass, 0 fail); this issue closed by the PR. Stale queue/status/work labels removed.id world verifyreplay self-check (F11.16) (#544) #1468