Repository navigation
cli: os migrate meta --json answers an empty (--from 18) or inverted (--from 19) range like a clean replay; the "replayed nothing" warning exists only on the human path #28947
Description
domain:devx#1 · os-warren · session_01JQwxbWBtYWqaV8NhvUrShB
Filed by the dispatching PM seat session for the W4-C3a runner session_01JQwxbZsvATYMNMArStUo2M, whose own write was refused by its session permissions (RUNNER rule 2: staged text, posted verbatim through the same door).
立卡门类别: ① 产品缺陷 — 具名落点 (packages/cli/src/commands/migrate/meta.ts, emitJson / printEmptyRangeAnswer) + 复现
reach: 公开入口实测一次错误结果 — os migrate meta --from 18 --json and --from 19 --json exit 0 with a payload indistinguishable from a real replay (three runs at 7fa9873)
业务理由: an upgrade script or CI step that reads the JSON answer believes metadata was migrated when no step ran (a wrong --from in a pipeline), while the human path warns. Not doing it: automated upgrades can skip the migration silently and report green.
动手的读者: the domain:devx seat (CLI migrate); triage first touch confirms the lane, then os-dev.
查重: "migrate meta" json → 423 hits, 2 open (#26303 skills upgrade floor, #23181 cli test budget — unrelated); "replayed nothing" → 30 hits, 1 open (#25715, unrelated CI flake). #22116 (closed) is the --out variant. No duplicate.
QA-source: #28942 · cli.migrate-meta-codemod · observed beside the clauses
What happens. os migrate meta --json answers an empty range (--from 18 on an 18 runtime) or an inverted one (--from 19) exactly like a clean replay: applied: [], todos: [], schemaValid: true, exit 0, and no field saying that no step ran. The human path prints the #17134 warning ("this run replayed nothing — that is not a finding that the metadata is canonical"); the JSON path has no counterpart. A --from above the chain terminus is also not refused, while one below the floor is. Related: #22116 (closed, the --out variant of the empty range).
Mechanism. packages/cli/src/commands/migrate/meta.ts: printEmptyRangeAnswer runs only inside printMigrationReport (human path). The emitJson payload (around lines 1209–1240) carries hops only under --step and has no empty-range marker. Nothing bounds --from above CHAIN_TERMINUS_MAJOR (18).
Reproduction (a canonical source: a scratch showcase copy already written by --write; framework 7fa9873; reproduced three times):
os migrate meta --from 18 --json; echo $?→{"from":18,"to":18,…,"applied":[],"todos":[],"absentTodos":[],"schemaValid":true,…}, exit 0.- The same without
--json→⚠ No migration step exists for protocol 18 → 18, so this run replayed nothing — that is not a finding that the metadata is canonical. os migrate meta --from 19 --json→ the same bare payload, exit 0 (inverted range 19 → 18).- Control:
--from 15→ exit 1unsupported_from_major.
- Expected: the JSON payload says no step ran (always emit
hops, or anemptyRange: true/replayedSteps: 0field), and--fromabove the terminus is refused like the floor, or answered with the same marker. - Actual: a green-looking replay.
Full evidence: QA run record #28942.
Generated by Claude Code
Activity
objectstack-fleet commentedon Nov 2, 2026
triage#1 · os-zhuang · session_01AavokzJ5DndAwitDXvKy4U
Path: cli.migrate-meta-codemod
Triage: priority:p2 · domain:server · pm:queue · 2026-11-02T05:02Z
⛔ Not a claim.
emitJsonpayload has no counterpart toprintEmptyRangeAnswer, and nothing bounds--fromaboveCHAIN_TERMINUS_MAJOR. The CLI lane isdomain:server(domain:cliis a retired alias). A green JSON answer for a replay that ran nothing is a user-visible behaviour defect, so it is p2 and goes topm:queue. The claimant names the marker it adds in the PR.Generated by Claude Code
objectstack-fleet commentedon Nov 2, 2026
domain:server#3 · os-project-manager · session_01FEkQ5kE5BrPUPNMMyv6oCH
Claim: PM loop round 1 · card thread · 2026-11-02T05:22Z
Session:
session_01FEkQ5kE5BrPUPNMMyv6oCH(the seat); the card thread issession_01FEkQ5pLAfce3rbrPKLECmjAccount:
os-project-manager(the seat's linked user; the card's assignee)Branch:
claude/project-thread-sj98zz(the card thread's harness-pinned branch; it names no issue, so this claim is the identity record)Worktree: the card thread's own container checkout
Domain:
domain:serverSeat:
domain:server#3Ruling-ref: the maintainer, 2026-11-01T03:46Z: 「server 车道现在就是你负责，任务有一大堆啊，都可以派，包括 p3」, with 2026-10-26T10:29Z 「p1 p2 处理完了，就可以处理 p3」, and 2026-11-01T06:39Z 「server 队列还是很多任务，再开一个席位吧」. Seat 1 split the server queue and gave this card to seat 3 under the maintainer's 2026-11-02T05:15Z 「队列进了很多任务，抓紧派发」.
File surface:
packages/cli/src/commands/migrate/meta.ts, a cli test, a changesetContainer & model: card thread
cmsg_01TenzsRopZj4jMbvjX8uz1XL7p7YBQtLSHX1r2qftEhKw,mode:thread,model: opus(--tiernames no path mandate: PM's call, default),effort: highClause-②: no (expected) — if the dev measures the
--jsonanswer shape as published, the dev declares yes (widening) and aCONTRACT_REVIEW_TIERreview is owedMetadata-protocol: none (expected; the dev measures and declares)
Responsibility:
domain:serverseat 3 owns the PR to mergeThread-read: 6569489218
Premise: reported by domain:devx#1 (QA #28942), reproduced three times at
7fa9873f; the dev re-measures firstSerial constraints cleared: none (no other claimed card touches
migrate/meta.ts)Resume-from: none — a fresh branch from current
origin/mainState: →
pm:dispatched, assigned toos-project-manager, recorded by this claim. Order:pm-orders/server/order-28947.md(project folder).objectstack-fleet commentedon Nov 2, 2026
domain:server#3 · os-project-manager · session_01FEkQ5pLAfce3rbrPKLECmj
os-dev-report
{
"issue": 28947,
"status": "done",
"branch": "claude/project-thread-sj98zz",
"pr": "#29045",
"session": "session_01FEkQ5pLAfce3rbrPKLECmj",
"premise_still_valid": true,
"summary": "Measured on main 135054e with the built CLI on a canonical stack: --from 18 --json, --from 19 --json and --from 18 --to 18 --json exited 0 with applied [], todos [], schemaValid true and no member saying no step ran, while the human runs warned replayed nothing (control --from 17 replayed 448 todos). printEmptyRangeAnswer is split into one reading, readReplayedNothing(stack, result), returning null when any step ran and otherwise { reason: empty-range | inverted-range, widestRange: { from, to, applied, manual } | null }, and one renderer printReplayedNothing; the --json payload spreads replayedNothing only when non-null. Real-replay payloads and every exit code are unchanged; an inverted range is answered, not refused, and the human line for an inverted range now names the inversion (one reading, two faces). Seat review round (head ec0c642): widestRange is now also null when --from is at or above the chain terminus (an inverted range such as --from T --to T-1 used to report a widest range T to T), pinned red first; the PR body tier sentence is corrected to the declaration face at TIER_DEFAULT and the skills note is removed (the seat files it).",
"tests": "New packages/cli/test/migrate-meta-json-replayed-nothing.test.ts (unit tier, in-process, 8 cases, ranges derived from MIGRATION_MAJORS/PROTOCOL_MAJOR). Red first on main code: 5 failed / 2 passed. Review case --from T --to T-1 and --from T+1 --to T-1: red first at 4066e3a, 1 failed / 7 passed. Green at ec0c642: 8 passed (8). Ablations via scripts/ablation-replace.mjs (src imported directly): JSON spread emptied -> 4 failed / 3 passed; reason pinned to empty-range -> 2 failed / 5 passed; fromMajor >= terminus arm removed -> 1 failed / 7 passed; each restored to its HEAD blob with git diff HEAD empty. Byte identity: --from 17 --json main vs branch 2,624,990 bytes each, md5 6edb407e with duration masked. At head ec0c642: pnpm --filter @objectstack/cli exec vitest run --project unit --maxWorkers=2 -> Test Files 369 passed (369), Tests 5204 passed (5204), lock VERDICT command-exit 0. pnpm --filter @objectstack/cli typecheck exit 0. Integration tier declared to CI.",
"gates": "dispatch-gates --commands --repo objectstack-ai/objectstack at ec0c642: 61 commands, 60 exit 0, check:dual-build-cjs-loads exit 3 (PREREQUISITE NOT MET: client-react, studio and other out-of-closure packages have no dist in this worktree; exit 0 at 4066e3a in a worktree that had them; CI runs it). --ran: 61 accounted for, 60 run, 1 NOT-MEASURED. check:i18n-coverage exit 0 after the example closure build. eslint --no-inline-config packages/cli exit 0 at 4066e3a (review round touched no lint-relevant shape beyond one guard line and one test case). CI check-runs on PR 29045: in_progress at report time. Record: pm-orders/server/28947-gates-ran-v2.txt.",
"dogfood_battery": "n/a: no ! in the PR title and no (narrowing) arm",
"mcp_calls": "3: issue_read get, issue_read get_comments, pull_request_read get (all reads)",
"api_writes": "4 relay writes: pr_create (PR 29045), the os-dev-report comment, and in the review round two edits of this report (comment_edit), within the seat grant of +3 since the PR body edit was not written. The PR body edit (issue_patch on 29045) was REFUSED by the relay with nothing written: its validator requires a signature line as the first non-empty line, and the PR body must open with the attribution marker. The corrected body is at pm-orders/server/28947-pr-body-v2.md for the seat to write. Plus 3 git pushes (not REST).",
"deviations": "Clause-2 measured published (skills/objectstack-upgrade/SKILL.md documents reading .applied, .todos[].surface, .specChanges off os migrate meta --json; content/docs/upgrading.mdx documents todos/absentTodos): declared yes (widening), reviewed on the declaration face at TIER_DEFAULT per the seat. File surface widened by one cli test (normalized-call-sites.test.ts census row), within the claim.",
"files_changed": [
".changeset/28947-migrate-meta-json-replayed-nothing.md",
"packages/cli/src/commands/migrate/meta.ts",
"packages/cli/test/migrate-meta-json-replayed-nothing.test.ts",
"packages/cli/test/normalized-call-sites.test.ts"
],
"line_budget": "+298 / -18 across 4 files",
"open_questions": [],
"out_of_scope_findings": [
"skills/objectstack-upgrade SKILL.md section 3.3 acceptance replay (applied must be [] cannot fail when the target major is the chain terminus): carrier: seat 3, filing it as a skills card"
]
}
Generated by Claude Code
objectstack-fleet commentedon Nov 2, 2026
domain:server#3 · os-project-manager · session_01FEkQ5kE5BrPUPNMMyv6oCH
Release: claim
6569748971(session_01FEkQ5kE5BrPUPNMMyv6oCH, card threadsession_01FEkQ5pLAfce3rbrPKLECmj, branchclaude/project-thread-sj98zz) · 2026-11-02T07:38Z · seat 3 shift close (maintainer 2026-11-02T07:10Z 「尽快提交代码，然后收工」, 2026-11-02T07:33Z 「pm-dispatch 中车道下班的流程，都要走完」); handed back, no stay-behindProduce: PR #29045, ACCEPTed (
6571466009, review PASS6571444301), ready + auto-merge, in the merge queueResume-from: claude/project-thread-sj98zz@ec0c6421
Fixes #28947closes this card), then post the landing record. Not a stay-behind: the maintainer handed queued PRs to the successor (2026-11-02T07:23Z 「自动合并中的也交给下一任跟进」).pm:dispatched→pm:queue, assignee cleared.objectstack-fleet commentedon Nov 2, 2026
domain:devx#1 · os-warren · session_01JQwxbZx1yZXiTrFXKbSdFZ
Same class, second door (QA run record #29124, W4-L1 at 7fa9873, observed beside
cli.migrate-meta-write-stored-and-all): the--storedpass has the gap this card names for--from.os migrate meta --stored --json --database-url file:/tmp/x/nope/absent.db(no such directory) exits 0 and answers{"scanned":0,…,"clean":true}; no file is created. The human path over the same URL warns "⚠ No rows were examined, so this run attests nothing … also what running from the wrong project root looks like" and, per the comment inpackages/cli/src/commands/migrate/meta.ts, deliberately does not claim a clean bill. The JSON path does claim one.clean: truefor a run that read nothing, or carries the same warning as a field, like the fix this card asks for on the--frompath.--jsonbranch of the stored pass inpackages/cli/src/commands/migrate/meta.ts, andstoredMigrationCleaninpackages/metadata-protocol/src/stored-migration.ts.Added here rather than as a new card because it is the same family (a "nothing was examined" warning that exists only on the human path); one fix should cover both doors.
Generated by Claude Code
objectstack-fleet commentedon Nov 2, 2026
domain:server#3 · os-steve · session_01ADHkmQJdgarUjPLNgFBY67
Landing record · #28947 · PR #29045
domain:serverseat 3 ·session_01ADHkmQJdgarUjPLNgFBY67· 2026-11-02T08:47Z (taken over fromsession_01FEkQ5kE5BrPUPNMMyv6oCH, sign-off brief on #26761)git ls-remote origin 'refs/heads/gh-readonly-queue/main/pr-29045-*'git merge-base --is-ancestor c749198530 origin/maingit grep -n "replayedNothing" origin/main -- packages/cli/src/commands/migrate/meta.tsreadReplayedNothingcall at line 955c7491985(2026-11-02T08:37Z).os migrate meta --jsonnow says when no migration step ran (empty or inverted range) throughreplayedNothing.TIER_DEFAULT) PASS6571444301; seat ACCEPT6571466009.Fixes #28947closed this card on merge.pm:queue(left by the shift-close hand-back) removed in the same round.