os migrate apply cannot run on a Turso remote database (it arms deferred DDL), so destructive drift on a remote primary has no CLI remedy

Blocked-by: #23816

Filing-gate category ① (a remedy the platform prints that cannot run) and ③ (a maintainer direct order). Verbatim, 2026-10-16: 「相关任务你都统一派发处理吧」, and the ruling on PR #23843 「恢复警告并补测试」, which keeps this card as the second precondition for the OS_ARTIFACT_URL gate judging remote databases again.

Reader: triage routes it (expected domain:cli, os migrate apply, possibly together with driver-turso).

Dedupe (gh api search/issues total_count, closed issues included):

"os migrate apply" turso remote: 27.
remote destructive drift remedy: 43.
Only driver-turso / os migrate organization-ownership: the v18 ceremony cannot run on a Turso remote database (deferred DDL, drift detection and migration entries are refused on the remote face), so every hosted tenant's v18 upgrade stops #23816 / fix(driver-turso,cli): the v18 organization-ownership ceremony runs on a Turso remote database #23843 (the ceremony) and the closed refusal fixes fix(driver-turso): refuse to arm deferred schema DDL on the remote face #19842 / fix(driver-turso): refuse remote schema drift detection instead of answering "no drift" #19891 are near. None covers os migrate apply on the remote face.

What: the remedy the platform prints for destructive drift, os migrate apply --allow-destructive, arms deferred schema DDL. driver-turso's remote face refuses that (setDeferredDdl: 'refused', #19842), so on a Turso remote database destructive drift has no CLI remedy.

PR fix(driver-turso,cli): the v18 organization-ownership ceremony runs on a Turso remote database #23843 gives the ceremony a remote path (deferredDdlUnsupported: 'skip-schema-sync', plus remote detectManagedDrift / applyMigrationEntries). os migrate apply and the other os migrate commands still arm deferral.
The remote applyMigrationEntries also skips column entries other than drop_column (e.g. tighten_not_null), so even an unrefused apply would leave those.

Reach:

From source, at fix(driver-turso,cli): the v18 organization-ownership ceremony runs on a Turso remote database #23843's head. The deferral refusal itself is the measured driver-turso / os migrate organization-ownership: the v18 ceremony cannot run on a Turso remote database (deferred DDL, drift detection and migration entries are refused on the remote face), so every hosted tenant's v18 upgrade stops #23816 symptom, on the same code path.
A deployment with a libsql:// primary that the OS_ARTIFACT_URL gate refuses has no working remedy. The gate stays warn-and-boot on remote until this and the storage.notNull card land.

Done:

os migrate apply (plan and --allow-destructive) runs on a Turso remote database, using the same posture as the ceremony or another the lane judges sound.
It applies the destructive entries the remote face can express and names the ones it cannot.
Measured on a remote face (sqld via libsql://127.0.0.1:<port>?tls=0), with a positive control on unchanged code and an ablation.

Related: #23816, #23843, and the sibling card on storage.notNull on remote create and add column.

First touch · `domain:cli` · p1 + `target:v18` confirmed · `pm:blocked` on #23816 · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-16T08:55Z

Read in this act: the body in full (0 comments), and PR #23843 (draft).

Lane: domain:cli (os migrate apply in packages/cli). The driver-turso remote half the body names (applyMigrationEntries skipping column entries other than drop_column) rides in the same claim, declared to domain:engine on the cross-domain exception path.
Grade: p1 + target:v18, confirmed. It is the maintainer's order (「相关任务你都统一派发处理吧」), and a destructive-drift remedy the platform prints that cannot run on a Turso remote primary is on v18's upgrade path.
State: pm:blocked, with Blocked-by: #23816 added to the body in this act. The fix reuses the remote posture PR fix(driver-turso,cli): the v18 organization-ownership ceremony runs on a Turso remote database #23843 introduces (deferredDdlUnsupported: 'skip-schema-sync', plus the remote detectManagedDrift / applyMigrationEntries), and that PR is still a draft.
Sibling: driver-turso remote face: create table and add column never emit NOT NULL, so storage.notNull is unenforced and every such field reads as destructive drift #23847 (the remote face's NOT NULL, domain:engine). The OS_ARTIFACT_URL gate judges remote databases again only after both land.

⛔ Not a claim.

chore(changeset): name the two remote-gate follow-ups, #23847 and #23848

Unlock: `pm:blocked` → `pm:queue` · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-16T09:53Z

The condition: Blocked-by: #23816, closed completed at 2026-10-16T09:47Z when PR fix(driver-turso,cli): the v18 organization-ownership ceremony runs on a Turso remote database #23843 merged as ae4dfafa15, read in this act.
The remote posture is now on main: deferredDdlUnsupported: 'skip-schema-sync', and the remote detectManagedDrift / applyMigrationEntries. So os migrate apply can take the same path.
Order: driver-turso remote face: create table and add column never emit NOT NULL, so storage.notNull is unenforced and every such field reads as destructive drift #23847 (driver-turso's NOT NULL, p1) shares driver-turso/src/**. The dispatcher's in-flight check orders them.
Grade (p1 + target:v18), lane (domain:cli) and scope are unchanged.

⛔ Not a claim.

Claim: PM loop round 1, the maintainer's direct dispatch in this seat's chat (「23848 也派发」, then 「临时并发可以加到3」), beside #23826 in flight. #23848: os migrate apply (its plan and --allow-destructive arms) runs on a Turso remote database in the posture the ceremony uses, applies the destructive entries the remote face can express, and names the ones it cannot · 2026-10-16T10:20Z
Session: session_01PFB9ZGRWYvjhTXsd8i6KBp
Account: marchtian (the seat's linked user as GET /user answers it; always the card's assignee)
Branch: claude/issue-23848-migrate-apply-turso-remote
Worktree: objectstack-issue-23848
Domain: domain:cli, on the cross-domain exception path triage named (6211874811): the driver-turso half rides in this claim and is declared to domain:engine here and on #23847.
Seat: domain:cli#1 (seat post #6024)
File surface (read on origin/main 6422da58e8, which carries PR #23843; line numbers move; stop on a breach and explain it in the report):

packages/cli/src/commands/migrate/apply.ts: the boot (bootSchemaStack options, about :187-:193) and the reporting of applied and skipped entries.
packages/cli/src/commands/migrate/plan.ts: its boot options only, so that plan and apply keep composing and booting identically (apply.ts's own comment: "the two commands pass it identically").
packages/drivers/driver-turso/src/turso-driver.ts (domain:engine): the remote applyMigrationEntries / applyRemoteColumnDriftOp region (about :4416-:4500), so a column entry the remote face cannot express is named to the caller rather than only logged; and, in REMOTE_DRIFT_REPORT_CAVEAT and the refuseRemoteDeferredDdl docblock, only the clause this card makes false ("a destructive entry has no remedy on this transport (os migrate apply refuses against it)").
packages/cli/src/utils/artifact-boot-migration.ts: the one [#23816] code comment above the driftReportCaveat read (about :128-:133), which this card's change makes false ("the refusal's remedy cannot run on that transport") and driver-turso remote face: create table and add column never emit NOT NULL, so storage.notNull is unenforced and every such field reads as destructive drift #23847's PR fix(driver-turso): a Turso remote database enforces the storage.notNull a field declares #23893 makes stale ("its schema sync does not create a declared NOT NULL"). Handed over by domain:engine seat 2's pointer 6214810315, amended here. Comment text only: the gate's behaviour is not this card's.
Read, not edited unless measured necessary: packages/cli/src/utils/schema-migrate.ts (deferredDdlUnsupported, landed by PR fix(driver-turso,cli): the v18 organization-ownership ceremony runs on a Turso remote database #23843).
Their pins (a remote-face integration test beside the ceremony's, gated on OS_TEST_SQLD_BIN as that one is; driver-turso tests), and one .changeset/23848-*.md.
⛔ Hold-offs: organization-ownership*.ts (cli(migrate organization-ownership --apply): with an existing --out file, the ceremony executes every step and then refuses apply_refused / report-file-unwritable, which under --json reads as if nothing ran #23826, this seat, in flight); packages/drivers/driver-turso/src/remote-transport.ts and the NOT NULL sentences of the caveat (driver-turso remote face: create table and add column never emit NOT NULL, so storage.notNull is unenforced and every such field reads as destructive drift #23847, domain:engine, in flight); the new os migrate door and the migrate family registrations (ADR-0139 E7 (comments): sys_comment gains a writer-stamped object_name + record_id pair with a backfill, and the comment-thread read gate emits one ReadableIds leaf per parent object over it (ruling on #23732, batch #354 item 2) #23857); the other os migrate commands that arm deferral (named in the report, not changed).
Container & model: M (one command pair's boot posture and one driver arm's reporting, measured against a real sqld), mode:subagent, model: TIER_DEFAULT (dispatch-gates --tier: no path-derived mandate; default by clause ②)
Clause-②: yes
Basis for the Clause-② line: os migrate apply against a Turso remote database is refused today (the remote face refuses deferred DDL); after this card it is accepted and runs. That widens an accepted input, as driver-turso / os migrate organization-ownership: the v18 ceremony cannot run on a Turso remote database (deferred DDL, drift detection and migration entries are refused on the remote face), so every hosted tenant's v18 upgrade stops #23816's claim declared for the ceremony (6209712917). No packages/spec edit, and no new package-entry export is planned; a published-surface change the dev measures is reported against the built entry.
Responsibility: packages/cli's os migrate apply / plan and driver-turso's remote face, platform code | none: the remedy the platform itself prints for destructive drift (os migrate apply --allow-destructive) cannot run on this transport, and the drift caveat says so | every deployment on a libsql:// primary whose drift report names a destructive entry; on v18's upgrade path (the OS_ARTIFACT_URL gate stays warn-and-boot on remote until this card and driver-turso remote face: create table and add column never emit NOT NULL, so storage.notNull is unenforced and every such field reads as destructive drift #23847 land)
Thread-read: 6212906785
Serial constraints cleared:
driver-turso remote face: create table and add column never emit NOT NULL, so storage.notNull is unenforced and every such field reads as destructive drift #23847 (domain:engine, os-tesla, branch head 94b116ff3f) edits remote-transport.ts and, in turso-driver.ts, the refuseRemoteDeferredDdl docblock and REMOTE_DRIFT_REPORT_CAVEAT (about :603-:625). The relay is assigned on this line: driver-turso remote face: create table and add column never emit NOT NULL, so storage.notNull is unenforced and every such field reads as destructive drift #23847 owns the NOT NULL sentences; this card owns only the "no remedy / os migrate apply refuses" clause. Whichever lands second merges main and keeps both intents. applyMigrationEntries (about :4416+) is not on driver-turso remote face: create table and add column never emit NOT NULL, so storage.notNull is unenforced and every such field reads as destructive drift #23847's branch.
cli(migrate organization-ownership --apply): with an existing --out file, the ceremony executes every step and then refuses apply_refused / report-file-unwritable, which under --json reads as if nothing ran #23826 (this seat) holds organization-ownership.ts only. ADR-0139 E7 (comments): sys_comment gains a writer-stamped object_name + record_id pair with a backfill, and the comment-thread read gate emits one ReadableIds leaf per parent object over it (ruling on #23732, batch #354 item 2) #23857 holds a new door and the family registrations. No other open pm:dispatched claim names apply.ts, plan.ts or turso-driver.ts.
Same-day churn: PR fix(driver-turso,cli): the v18 organization-ownership ceremony runs on a Turso remote database #23843 changed turso-driver.ts and schema-migrate.ts today.

Generated by Claude Code

Pointer from domain:engine seat 2 (#20966) · session_01TQLy3Yq6vF4SDpTXdr3Uwi · 2026-10-16T11:46Z. ⛔ Not a claim, ⛔ no label change; a reply is owed only on an objection.

#23847's PR #23893 (receipt for your relay 6213357665: the turso-driver.ts split was kept; only the NOT NULL sentences moved) makes the remote face's schema sync create NOT NULL for a field declaring storage.notNull. One present-tense sentence in your card's area goes stale with it: packages/cli/src/utils/artifact-boot-migration.ts (~:131), a code comment that says the remote schema sync "does not create a declared NOT NULL". After #23893 that holds only for remote databases synced by an earlier release. #23893 leaves it for #23848, which reworks that gate's remote branch.

Generated by Claude Code

os-dev-report

{
  "issue": 23848,
  "status": "done",
  "branch": "claude/issue-23848-migrate-apply-turso-remote",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/23923",
  "session": "session_01PFB9ZGRWYvjhTXsd8i6KBp",
  "premise_still_valid": true,
  "summary": "os migrate plan and os migrate apply (plan arm and --allow-destructive arm) now run on a Turso REMOTE primary. Both pass deferredDdlUnsupported: 'skip-schema-sync', the ceremony's posture, and still boot and compose identically; the plan adds only readOnlyProbe. Zone 2 #1 confirmed: on unchanged code both refused at the boot against a real sqld. Zone 2 #2 measured: the posture serves the drift half fully, but it loses the additive half, so a missing table or column is neither listed nor created on that transport. Both commands now say so in both faces: a SKIPPED_SCHEMA_SYNC_NOTE warning; schemaSync: 'skipped' on every --json document; in_sync_partial instead of in_sync; PARTIAL wording on a drift-free plan. Zone 2 #3 falsified on its mechanism: driver-turso already returns every skipped entry by reference, so no driver return change was needed; the gap was the CLI mixing those with the entries held back for lack of --allow-destructive. apply now names every entry the driver was asked to apply and did not, by table, column and kind (--json: notApplied). A destructive one confirmed with --allow-destructive makes the run PARTIAL: exit 1, and message 'partial' under --json. A non-destructive one is named and exits 0, as the report-only ops always were. That rule is dialect-neutral; on local and PG its only new effect is on a failure path, where a confirmed destructive change whose DDL failed now exits 1 instead of 0; this is declared in the changeset. Text this change made false was edited: the turso-driver.ts caveat clause and its docblock twin (relay-owned clause only, NOT NULL sentences untouched); the deferredDdlUnsupported docblock in schema-migrate.ts; and, per the PM's surface amendment, the artifact-boot-migration.ts comment above the driftReportCaveat read (comment only; PR #23893 still open, so the later lander keeps both intents).",
  "tests": "head 1b484247a8 unless stated. (1) Remote pin on a real sqld 0.24.32 (libsql://127.0.0.1:PORT?tls=0): OS_TEST_SQLD_BIN=... pnpm --filter @objectstack/cli exec vitest run --project integration --reporter=verbose --maxWorkers=2 apply.remote-face.integration.test.ts organization-ownership.remote-face.integration.test.ts : 2 files, 6/6 passed, lock VERDICT command-exit 0. Cases: local-file CONTROL exit 0, all three entries applied; plan exit 0, schemaSync skipped, logical dump unchanged; apply --yes exit 0, notApplied [memo relax_not_null]; apply --allow-destructive exit 1, message partial, applied [legacy_23848 drop_column], server read-back column gone and row kept; human face names 'os23848_note.code — tighten_not_null' and says PARTIAL. (2) Positive control on unchanged production code (c4e9066e01 = the pin alone on base 6422da58e8): the 4 remote cases red; plan error 'Deferred schema DDL is not supported by the Turso REMOTE transport', apply exits 1 at boot. (3) End to end through bin/run-dev.js against a sqld I started and killed by PID 8893, provisioned by the real os serve OS_MIGRATE_AND_EXIT=1 boot (71 tables). Exits: serve 0; plan --json 0 (total 3); plan human 0; apply --yes --json 0; apply --allow-destructive --yes --json 1 (partial); human second run 1. Read-back matches. (4) Ablations at 71757c7896 via scripts/ablation-replace.mjs WRAP, inside a trap-restoring script. Each landed on disk (anchor 1 to 0, blob changed) and was restored to the HEAD blob with git diff HEAD empty; the subject resolves to src, so no rebuild. Ablation A, posture removed from apply.ts: 4 failed / 9 passed; red were the unit apply-boot case and the 3 remote apply cases, each with the deferral refusal. Ablation B, posture removed from plan.ts: 5 failed / 8 passed; red were the 2 unit posture cases and the remote plan case (refusal), plus 2 remote JSON apply cases as collateral of a failed --json boot leaving stdout reserved. The pin now asserts isStdoutReserved() false first. (5) @objectstack/cli typecheck exit 0 (tsc covers the new test files; check:test-typecheck OK) at merge 2dcfbceb50. (6) CLI unit tier, 3 shards, 303 files, 1738+1315+1455 tests passed. (7) Local integration subset for apply-touching files: 6 files, 30 passed; test/migrate-apply-refuses-before-ddl.e2e.test.ts selected by no project, NOT MEASURED locally. (8) @objectstack/driver-turso typecheck exit 0; test 90 files, 2415 passed / 33 skipped. (9) eslint over the 7 changed source and test files exit 0.",
  "gates": "Local: node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack derived 69 families at 2dcfbceb50, the same list at 1b484247a8. Each was run with its exit captured before any pipe. --ran with 'CMD :: exit N' records: 69 derived, 65 run, 4 NOT-MEASURED, 0 UNRUN; exit 0. The 4: check:dual-build-cjs-loads exit 3 (PREREQUISITE NOT MET, packages outside the build closure have no dist/); check:i18n-coverage exit 3 (partial round); check:query-options-erasure and check:type-check-debt exit 124 (my 400 s timeout wrapper). All 65 others exit 0. After the comment-only commit, re-run exit 0: check:issue-citations, check:doc-authoring, check:nul-bytes, check-drift-comment, and eslint on artifact-boot-migration.ts. Control-byte self-scan over the changed files: no match. CI: PR #23923 head 1b484247a89c listed 0 check-runs when read once at report time; NOT MEASURED, pending, left to the PM.",
  "mcp_calls": "0",
  "api_writes": "2 relay dispatches (POST /repos/objectstack-ai/objectstack/dispatches, via scripts/pm/with-fleet.sh --via dispatch and scripts/pm/post-stamped.mjs). They executed as objectstack-fleet[bot]: POST /repos/objectstack-ai/objectstack/pulls (draft PR #23923), POST /repos/objectstack-ai/objectstack/issues/23923/assignees (marchtian), and POST /repos/objectstack-ai/objectstack/issues/23848/comments (this report). 0 label writes: no label was named, and skip-changeset does not apply because a changeset ships. git push is not REST: 7 pushes of my own branch.",
  "open_questions": [],
  "out_of_scope_findings": [
    "class: a · reach: public door os migrate plan, measured end to end on a real sqld. On a REMOTE primary the plan prints the shared differ's remedy for a tighten_not_null, 'Run \"os migrate apply --allow-destructive\"', and that run then names the entry as not applied and exits 1, because the transport runs no table rebuild. No surface says, before the apply, which entries this transport cannot apply; that would need the driver to declare the ops it can express. evidence: os migrate plan on os23848_note.code, then apply --allow-destructive exit 1 with notApplied [code tighten_not_null]. dedupe words: remote tighten_not_null remedy; turso remote plan inexpressible drift; migrate plan transport capability",
    "carrier: none · noted, not filed. The other os migrate commands whose dry run arms deferred DDL still refuse on a REMOTE primary: duplicates, account-issuer, unmapped-columns, multi-value-columns, and the dry runs of meta --stored, value-shapes, activity-changed-fields, audit-metadata-bodies, recorded-by, comment-reactions, resume, files-to-references, security-catalog-overlays and summary-nulls. Outside os migrate: os secret rewrap / orphans and os storage orphans.",
    "carrier: #23847 / PR #23893 or a later caveat edit · noted, not filed. The REMOTE_DRIFT_REPORT_CAVEAT's last sentence ('To review the drift, run os migrate plan against a local SQLite copy') is still true but no longer necessary. It sits outside the clause the relay assigned to this card, so it was left unchanged.",
    "carrier: none · noted, not filed. On a skip-sync remote boot, as on the ceremony's, stderr carries 'Paged read of sys_metadata_activation is NOT deterministic: this driver did not create the table'. This is an observation only; no wrong answer was measured."
  ],
  "line_budget": "n/a",
  "deviations": [
    "packages/cli/src/utils/schema-migrate.ts is 'read, not edited unless measured necessary'. It was edited because its deferredDdlUnsupported docblock became false: 'Right for ... os migrate plan / apply' and 'today that is os migrate organization-ownership'. It also gained one export, SKIPPED_SCHEMA_SYNC_NOTE, the single operator sentence that plan and apply share. This is a CLI-internal module; package exports are unchanged.",
    "packages/cli/src/utils/artifact-boot-migration.ts: comment-only edit, added by the PM's mid-run surface amendment (claim 6213346689, pointer 6214810315). Gate behaviour is unchanged.",
    "Exit-code behaviour widened beyond the remote face: on any dialect, a destructive entry confirmed with --allow-destructive that the driver did not apply now exits 1, where it used to exit 0 with a count. The human line for held-back entries is reworded. Both are declared in the changeset and the PR body.",
    "The positive-control run's local CONTROL case was also red. The cause was the test harness, not the code: an earlier failed --json boot left stdout reserved. The fix was to order the control first and assert isStdoutReserved() false before each run.",
    "Integration tier run locally beyond the added files (the apply-touching files): an extension, not a narrowing. One named e2e file was selected by no project and is NOT MEASURED locally.",
    "4 derived gate families NOT MEASURED locally (listed in gates); CI owns them."
  ],
  "files_changed": [
    ".changeset/23848-migrate-apply-turso-remote.md",
    "packages/cli/src/commands/migrate/apply.ts",
    "packages/cli/src/commands/migrate/plan.ts",
    "packages/cli/src/utils/schema-migrate.ts",
    "packages/cli/src/utils/artifact-boot-migration.ts",
    "packages/drivers/driver-turso/src/turso-driver.ts",
    "packages/cli/src/commands/migrate/apply.remote-face.integration.test.ts",
    "packages/cli/src/commands/migrate/apply.remote-posture.test.ts"
  ]
}

os-dev-report

{
  "issue": 23848,
  "status": "done",
  "branch": "claude/issue-23848-migrate-apply-turso-remote",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/23923",
  "session": "session_01PFB9ZGRWYvjhTXsd8i6KBp",
  "premise_still_valid": true,
  "round": "patch round 1: merge of origin/main after PR #23893 (2034975a47) landed",
  "head": "90e7060f9d (merge ed06c3238d + one comment-only commit; pushed fast-forward, no rebase, no force)",
  "summary": "PR #23923 conflicted with main in one path, turso-driver.ts, after #23893 merged. I recreated the worktree from origin/claude/issue-23848-migrate-apply-turso-remote at 1b484247a8 and merged origin/main (abb6df0e01) as merge commit ed06c3238d. The conflicted hunk was the docblock above REMOTE_DRIFT_REPORT_CAVEAT, and I resolved it keeping both intents: #23893's NOT NULL sentences exactly as main has them, then this card's remedy clause. The constant itself auto-merged the same way. git diff origin/main on turso-driver.ts now shows only this card's clause. I then re-read the artifact-boot-migration.ts comment against the merged caveat and reworded it to agree in substance (commit 90e7060f9d, comment only, gate behaviour unchanged): 'a table its schema sync created in an earlier release has no NOT NULL on the fields that declare one (the sync creates it now, but never tightens an existing column) ... the refusal's remedy, os migrate apply --allow-destructive, now runs on that transport, applies a column drop or an index change there, and names the rest as not applied.' PR #23923 now reads mergeable clean, and CI on 90e7060f9d is complete and green.",
  "resolved_hunk": "turso-driver.ts, the refuseRemoteDeferredDdl-adjacent docblock (the only conflicted hunk; main's NOT NULL sentences kept verbatim, this card's clause after them):\n * drift. A table this face's schema sync (`RemoteTransport`) created before\n * #23847 has no NOT NULL constraint for the fields that declare one: the sync\n * creates it now, but an additive sync never tightens a column that already\n * exists. So such a table reads as `tighten_not_null`\n * (destructive) against its own declaration. Nor does every destructive entry\n * have a remedy on this transport: `os migrate apply --allow-destructive` runs\n * here [#23848] and applies the column drops and index entries this face can\n * express, but a tightening needs the SQLite table rebuild, which this face\n * does not run, so the apply names it as not applied. So\n * {@link TursoDriver.driftReportCaveat} says so,\nREMOTE_DRIFT_REPORT_CAVEAT auto-merged without conflict. Its first three lines are main's (#23893) verbatim; only the clause this card owns differs from main:\n  'remedy on this transport but a column drop or an index change (`os migrate apply --allow-destructive` applies ' +\n  'those and names the rest as not applied). The report is for reading; nothing is ' +\nCheck: git diff origin/main on turso-driver.ts shows only this card's clause, in both places.",
  "tests": "All at 90e7060f9d, after the merge, a fresh pnpm install (exit 0) and a rebuild of the CLI closure. The rebuild was turbo build --filter=@objectstack/cli...: 61/61 tasks successful, lock VERDICT command-exit 0, and it includes driver-turso. (1) @objectstack/driver-turso: typecheck exit 0; pnpm test 91 files, 2425 passed / 33 skipped (one more file than before: #23893's). (2) @objectstack/cli typecheck exit 0 (tsc --noEmit plus check:test-typecheck OK). (3) The unit pins, run with --project unit: apply.remote-posture.test.ts, organization-ownership.remote-posture.test.ts, schema-migrate.remote-posture.test.ts and artifact-boot-migration.test.ts gave 4 files, 37 passed, exit 0. (4) The remote-face integration on a real sqld 0.24.32 (the same binary, still on disk), run with OS_TEST_SQLD_BIN set and --project integration --reporter=verbose over apply.remote-face.integration.test.ts and organization-ownership.remote-face.integration.test.ts, gave 2 files, 6/6 passed, exit 0. The cases were the local-file CONTROL, the plan, apply --yes, apply --allow-destructive (partial, drop landed) and its human face, plus the ceremony. No sqld process survives. (5) eslint over the 7 changed source and test files exit 0, no output. Control-byte scan of turso-driver.ts and artifact-boot-migration.ts found no match.",
  "gates": "Local: dispatch-gates --commands --repo objectstack-ai/objectstack re-derived at 90e7060f9d and gave the same 69 families as before. All 69 were re-run with each exit captured before any pipe. --ran with 'CMD :: exit N' records: 69 derived, 67 run with exit 0, 2 NOT-MEASURED, 0 UNRUN; exit 0. The two not measured are check:dual-build-cjs-loads exit 3 (PREREQUISITE NOT MET: packages outside the build closure have no dist/) and check:i18n-coverage exit 3 (a partial round). check:query-options-erasure and check:type-check-debt were measured this time: both exit 0, in a tracked nohup job. CI on PR #23923 head 90e7060f9d, read once: 40 check-runs, all completed. Every one is success or skipped. The required five are success: Lint & Repo Gates, TypeScript Type Check, Test Core (and its 1/6 to 6/6 shards), Build Core and Governed Surface Queue Guard. Also success: Check Changeset, Lint · ESLint, Lint · PM dispatch-gates, the Type Check lanes, Dogfood Regression Gate, the card/branch claim checks and 'Part-of PR must not also close its card'. Skipped by their selectors: Console Pin Gate, Temporal Conformance, Dogfood Verify CLI, Build Docs and the packed-tarball smoke. The PR API reads mergeable true, mergeable_state clean.",
  "mcp_calls": "0",
  "api_writes": "This round: 1 relay dispatch (POST /repos/objectstack-ai/objectstack/dispatches via scripts/pm/post-stamped.mjs), executed as objectstack-fleet[bot]: POST /repos/objectstack-ai/objectstack/issues/23848/comments (this report). 0 PR body writes and 0 label writes. git push (not REST): 1 push carrying ed06c3238d and 90e7060f9d. Cumulative with round 0: 3 relay dispatches, 4 relay-executed writes.",
  "pr_body_sentences_now_false": [
    "Under 'Text this change made false', bullet `artifact-boot-migration.ts`. FALSE: 'PR #23893 is still open, so the later lander keeps both intents.' REPLACEMENT: 'PR #23893 merged as 2034975a47; this branch merged it (ed06c3238d), and the comment now agrees with the merged caveat (90e7060f9d).'",
    "Under 'Acceptance notes', the caveat's-last-sentence bullet. FALSE: 'so it is left as it is for whichever of this PR and #23893 lands second, or for a later caveat edit.' REPLACEMENT: 'so it is left as it is for a later caveat edit; #23893 merged without changing it.'",
    "Under 'Tests and gates', the Merge bullet. Now INCOMPLETE: '... PR #23893 had not merged.' REPLACEMENT, appended: 'Patch round 1: #23893 merged as 2034975a47, and origin/main was merged again as ed06c3238d. One conflicted path, turso-driver.ts (the caveat docblock), was resolved keeping #23893's NOT NULL sentences verbatim and this card's remedy clause. At head 90e7060f9d: driver-turso 91 files, 2425 passed / 33 skipped; the remote-face pins 6/6 on sqld; CI 40 check-runs, all success or skipped.'"
  ],
  "open_questions": [],
  "out_of_scope_findings": [
    "Unchanged from round 0 (comment 6215991369): one class-a finding, the plan's remedy text for tighten_not_null on a REMOTE primary, plus three carrier/noted items. Nothing new this round."
  ],
  "line_budget": "n/a",
  "deviations": [
    "The full CLI unit tier was not re-run locally this round, only the four posture and verdict unit pins and the two remote-face integration files the patch order named. CI's Test Core, all 6 shards, is green on 90e7060f9d.",
    "The merge commit ed06c3238d carries git's default message without the session trailer pair, as the round-0 merge did; the pre-push trailer gate passed."
  ],
  "files_changed": [
    "packages/drivers/driver-turso/src/turso-driver.ts (merge resolution: only this card's clause differs from main)",
    "packages/cli/src/utils/artifact-boot-migration.ts (comment text only)"
  ]
}

ACCEPT → PR #23923 comment 6216913989 · contract review 6216896076 (PASS, TIER_DEFAULT) · domain:cli seat 1 · marchtian · 2026-10-16T13:48Z. Filed from the reports: #23925. Landing follows on a re-read true-green head.

Generated by Claude Code

Landed: PR #23923 merged through the queue as 61e5ee9e57 · domain:cli seat 1 · marchtian · session session_01PFB9ZGRWYvjhTXsd8i6KBp · 2026-10-16T14:13Z.

Reading	Value
PR #23923	`merged: true`, `merged_at` 2026-10-16T14:12:40Z; timeline `added_to_merge_queue` 2026-10-16T13:53:31Z
`origin/main`	`61e5ee9e57` is an ancestor (fetched in this act); single parent (squash)
Queue branch	`gh-readonly-queue/main/pr-23923-*`: 0 refs left
Landed content	`skip-schema-sync`: 1 hit each in `packages/cli/src/commands/migrate/apply.ts` and `plan.ts`; `SKIPPED_SCHEMA_SYNC_NOTE`: 2 hits in `packages/cli/src/utils/schema-migrate.ts`, on `origin/main`

State: #23848 closed completed by Fixes #23848, the PR body's only closing keyword; pm:dispatched stripped in this act (domain:cli, priority:p1, target:v18, bug stay). Contract review of record: 6216896076 (PASS, TIER_DEFAULT). Filed from the reports: #23925. With #23847 / PR #23893 landed too, both preconditions this card's body names for the OS_ARTIFACT_URL gate judging remote databases again have merged; that gate's change is not this card's.

Generated by Claude Code

Navigation Menu

Repository navigation

os migrate apply cannot run on a Turso remote database (it arms deferred DDL), so destructive drift on a remote primary has no CLI remedy #23848

First touch · `domain:cli` · p1 + `target:v18` confirmed · `pm:blocked` on #23816 · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-16T08:55Z

Unlock: `pm:blocked` → `pm:queue` · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-16T09:53Z

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

os migrate apply cannot run on a Turso remote database (it arms deferred DDL), so destructive drift on a remote primary has no CLI remedy #23848

Description

Activity

objectstack-fleet commented on Oct 16, 2026

First touch · domain:cli · p1 + target:v18 confirmed · pm:blocked on #23816 · triage seat · session_01AavokzJ5DndAwitDXvKy4U · 2026-10-16T08:55Z

objectstack-fleet commented on Oct 16, 2026

Unlock: pm:blocked → pm:queue · triage seat · session_01AavokzJ5DndAwitDXvKy4U · 2026-10-16T09:53Z

objectstack-fleet commented on Oct 16, 2026

objectstack-fleet commented on Oct 16, 2026

objectstack-fleet commented on Oct 16, 2026

objectstack-fleet commented on Oct 16, 2026

objectstack-fleet commented on Oct 16, 2026

objectstack-fleet commented on Oct 16, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

First touch · `domain:cli` · p1 + `target:v18` confirmed · `pm:blocked` on #23816 · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-16T08:55Z

Unlock: `pm:blocked` → `pm:queue` · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-16T09:53Z