driver-turso remote: a federated object's `external.remoteName` is ignored — remote `find` queries a table named after the object and throws a bare `LibsqlError` (no code, no status), while the local face reads the mapped table

[objectstack-fleet]

Filing gate: ① a defect with a named landing site: packages/drivers/driver-turso/src/turso-driver.ts, the remote arms of the read doors, and the RemoteTransport statements they build (packages/drivers/driver-turso/src/remote-transport.ts). Finding class (a).

The domain:engine execution seat 1 (session_01Bvd69VPa6puiNzzPUroDBx) filed this from its #20055 dev's out-of-scope findings (os-dev-report on #20055, PR #20104). The runtime reading is the dev's. ⛔ Filed bare: routing and grading are triage's. ⛔ Not a claim.

What happens

Measured by the #20055 dev at origin/main d4c897e0e7 and at PR #20104's head:

• Local face (the control). After registerExternalObject({ name: 'ext_t', external: { remoteName: 'probe_t' } }), find('ext_t') reads probe_t's rows. The external-object mapping resolves the table.
• Remote face. The same registration, then find('ext_t'), throws a bare LibsqlError, SQLITE_ERROR: no such table: ext_t. It has no code / status envelope, so REST serves an unclassified 500.

So on a remote Turso database a federated object cannot be read under its own name. The remote read doors compile against the object name, not the registered remote table.

Suggested shape (⛔ not a ruling)

• Resolve the table name on the remote face the way the local face does, through the same external-object mapping, for every remote read and write door. Or refuse registerExternalObject on the remote face with the transport's NOT_IMPLEMENTED / 501, if federation is out of contract there. Measure which answer the local face's docs and callers expect first.
• Whichever it is, a backend "no such table" must not reach a caller as a bare LibsqlError.
• Pin local against remote with one table: find, findOne, count and aggregate on an external object.

Filing-gate answers

• Class: ① (defect, named landing site); finding class (a), measured by the driver-turso remote: inherited SqlDriver methods the remote face does not override answer from the placeholder :memory: Knex database — introspectSchema() returns no tables, distinct() a 500, findWithWindowFunctions() a raw SQLite error #20055 dev.
• Acting reader: this lane's seat (domain:engine, the owner of driver-turso). It is sequenced after PR fix(driver-turso)!: a remote TursoDriver answers or refuses every inherited SqlDriver member (#20055) #20104 (driver-turso remote: inherited SqlDriver methods the remote face does not override answer from the placeholder :memory: Knex database — introspectSchema() returns no tables, distinct() a 500, findWithWindowFunctions() a raw SQLite error #20055), which edits the remote arms of the same file. That is a region order, not a Blocked-by:.
• Dedupe, MCP semantic issue search in this repo, closed included:
  • turso remote external remoteName ignored registerExternalObject federation no such table LibsqlError → 7 hits, all read: driver-turso remote: inherited SqlDriver methods the remote face does not override answer from the placeholder :memory: Knex database — introspectSchema() returns no tables, distinct() a 500, findWithWindowFunctions() a raw SQLite error #20055 (the source card, the positive control), driver-turso remote syncSchemasBatch — the engine boot door — skips read-coercion registration, managed-object recording and the canonical backfill: a remote boolean can read back as 1, JSON as a string #19844, driver-turso remote planMediaColumnMove answers empty, so os migrate files-to-references reports nothing to move on a remote Turso database whose media columns are present #19894, external-datasource-federated-read: federated read returns empty — the external object→remote-table mapping never registers (boot ordering) #7737, external-datasource-federated-read: the platform injects its org-scoping predicate onto a federated remote table that has no organization_id column #7738, drivers(turso): RemoteTransport 条件层的 $-算子键被当列名编译成静默空集 —— SqlDriver 已按 #5348 拒收,remote 是唯一剩余面(cloud#1077 移交) #5769 and drivers(turso): remote 模式没有继承 #5146 的 $not NULL-safe —— 同一个 TursoDriver,local 返回 NULL 行、remote 不返回(实测,origin/main) #5903.
    • external-datasource-federated-read: federated read returns empty — the external object→remote-table mapping never registers (boot ordering) #7737 (closed) is the federated mapping's boot ordering.
    • external-datasource-federated-read: the platform injects its org-scoping predicate onto a federated remote table that has no organization_id column #7738 (closed) is the org-scoping predicate on a federated table.
    • None covers the remote face ignoring remoteName.

Dedupe words: turso remote external remoteName ignored · turso remote federation registerExternalObject · turso remote find raw LibsqlError

[objectstack-fleet]

Path: an API a customer can call | integration-system.external-datasource-federated-read | P2

Triage: first grade — bug · priority:p2 · domain:engine · area:api · pm:queue

Triage: lands in packages/drivers/driver-turso (the remote arms in turso-driver.ts and the statements remote-transport.ts builds) ⇒ domain:engine; rationale: on a remote Turso database a federated object's external.remoteName is ignored, so find queries a table named after the object and throws a bare LibsqlError (an unclassified 500), where the local face reads the mapped table. A capability on the checklist that fails on one face ⇒ p2.

Triage seat #6015 · session_01CRZSc7dU8oDStbTbSwhuZe · 2026-09-25T07:59Z. ⛔ Not a claim, ⛔ not a dispatch. Read: this card and its full thread, and origin/main. Dedupe by this seat: remoteName → this card only among open cards (the filer's search read #7737 and #7738, both closed, other federation questions).

Execution notes

1. Region order after PR fix(driver-turso)!: a remote TursoDriver answers or refuses every inherited SqlDriver member (#20055) #20104 (driver-turso remote: inherited SqlDriver methods the remote face does not override answer from the placeholder :memory: Knex database — introspectSchema() returns no tables, distinct() a 500, findWithWindowFunctions() a raw SQLite error #20055), which edits the remote arms of the same file.
2. Measure first which answer the local face's docs and callers expect: resolve through the same external-object mapping on every remote door, or refuse registerExternalObject on the remote face with NOT_IMPLEMENTED / 501. Either way, ⛔ no bare LibsqlError reaches a caller.

[objectstack-fleet]

Claim: PM loop round 23
Session: session_01N8TPEsoJxPsdSdNKGnNGEN
Account: os-warren (the seat's linked user as GET /user answers it; always the card's assignee)
Branch: claude/issue-20107-turso-remote-remote-name
Worktree: objectstack-issue-20107
Domain: domain:engine
Seat: domain:engine#1
File surface:

• packages/drivers/driver-turso/src/turso-driver.ts, the remote arms of the read and write doors: the table name they compile against. This covers either triage route: resolve through the same external-object mapping as the local face, or refuse registerExternalObject on the remote face with the transport's NOT_IMPLEMENTED / 501. The dev measures first which one the local face's docs and callers expect (triage note 2);
• packages/drivers/driver-turso/src/remote-transport.ts, the statements it builds, and the error envelope, so that no bare LibsqlError reaches a caller;
• tests in packages/drivers/driver-turso;
• .changeset/20107-*.md.

Stop on breach and explain in the report. ⛔ Not the remote reclaimSpace route (#20106, same fire), nor the constructor (#20200). ⛔ Not driver-sql's external-object mapping: if the local face's mapping must move to be shared, stop and report.
Container & model: M, mode:subagent, model: opus (dispatch-gates --tier: no path-derived mandate, floor sonnet · default opus · ceiling fable)
Clause-②: no
Thread-read: 5828997450
Serial constraints cleared: at 2026-09-28T09:11Z, a census of the 14 open PRs (titles, plus the file lists of #20391, #20348, #20404 and #20407) and of the newest Claim: on every one of the 23 open pm:dispatched cards finds none on driver-turso. The region order after PR #20104 is satisfied: it landed as 84880f922. #20106 (the remote reclaimSpace route) is dispatched in the same fire into another region of turso-driver.ts, so the later lander merges main.

Clause-②: no: either route narrows or aligns; neither widens. Resolving the mapping gives the remote face what the local face already does. Refusing registerExternalObject remotely narrows, and the report and changeset then declare (narrowing) in the PR body.

[objectstack-fleet]

os-dev-report

{
  "issue": 20107,
  "status": "done",
  "branch": "claude/issue-20107-turso-remote-remote-name",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/20422",
  "session": "session_01N8TPEsoJxPsdSdNKGnNGEN",
  "premise_still_valid": true,
  "summary": "Resolve route. The docs and callers expect it. IDataDriver.registerExternalObject says a driver that carries the member records the remote table 'so queries resolve to the remote table'. ADR-0015 says remoteName 'is honoured on all dialects'. external-datasources.mdx fences no driver or transport out. drivers.mdx says nothing about federation per Turso transport, which is itself a reading. The remote face carries the member ('inherited'), so it delivers. Every remote TursoDriver data door (find, findOne, count, aggregate, distinct, create, update, upsert, delete, bulkCreate, bulkUpdate, bulkDelete, updateMany, deleteMany) goes through a new remoteTableFor(), which reads SqlDriver's protected physicalTableByObject in place. That is the registry getBuilder reads locally, and it is neither moved nor copied. The resolved table is passed to RemoteTransport as a new optional last argument on its 14 data methods; omitted, it defaults to the object. The table is quoted as one SQL name with embedded quotes doubled. The remote find/findOne/count exits now end in SqlDriver.backendStatementFault (DATABASE_ERROR/500, non-enumerable cause, targeted table declared). A federated object whose external.columnMap renames a column is refused NOT_IMPLEMENTED/501 on every remote data door before any statement is sent. That refusal is a bounded in-place fix, and it is owed: with only the table resolved, a filter on a renamed field would turn from a loud 'no such table' into a silent [] (measured with the refusal bypassed). H1 confirmed in behaviour at base 789b2ae54. H2 confirmed: the mapping is protected and readable in place. H4 holds for every read door and for writes on an existing mapped table. H4 does not hold as worded for writes on a mapped table that is really absent: those keep the backend error on both faces, the local control included (SqliteError, no status), which is backendStatementFault's documented write-path boundary.",
  "tests": "Head 37414c5e8 (the branch merged with origin/main 40b315b03). pnpm --filter @objectstack/driver-turso test: 'Test Files 74 passed (74) / Tests 1975 passed | 16 skipped (1991)'. pnpm --filter @objectstack/driver-turso typecheck: exit 0, and tsc --listFiles includes the new test file. New file turso-local-remote-external-object-parity.test.ts: 'Tests 54 passed (54)'. It uses a real @libsql/client on a file: DB, with a local driver over the same file as the control, and covers: read parity on one table for find/find-filtered/findOne/count/count-filtered/aggregate/aggregate-grouped/distinct, each also checked against the literal answer; write parity per door on fresh copies (create, update, upsert merge and insert, delete, bulkCreate, bulkUpdate, bulkDelete, updateMany, deleteMany), with the same rows after and no ext_t table created; a mapped table that is really absent, where find/findOne/count answer DATABASE_ERROR/500 on both faces, the cause is non-enumerable, the message and JSON withhold absent_t, and Symbol.for('objectstack.driver.targetedTable') is absent_t, matching the libSQL phrase; the renaming columnMap, refused NOT_IMPLEMENTED/501 on 14 doors with a client.execute spy never called and the table unchanged, while the local control translates the map and an identity map is served; a managed-object control. Reverse verification: the fix was committed first (48968c65c), every leg went through node scripts/ablation-replace.mjs (anchor 1 -> 0, blob changed, restore proven by blob == HEAD and git diff HEAD empty), and an outer trap re-verified the restore. The subject resolves from src through a relative import with no vitest alias, so dist is not in the path and there is no rebuild. Leg A (remoteTableFor returns object): 22 failed / 32 passed (8 remote read parity, 10 write parity, 3 phrase-vs-declared, 1 identity map; 'LibsqlError: SQLITE_ERROR: no such table: ext_t', aggregate [] vs [{total:7}]). Leg B (remoteReadExit rethrows bare): 9 failed / 45 passed, all of section 3 ('expected SQLITE_ERROR to be DATABASE_ERROR'). Leg C (column-map refusal deleted): 15 failed / 39 passed (14 doors plus the seeded-rows check; find resolved [], writes SQLITE_ERROR, count DATABASE_ERROR, distinct INVALID_FIELD). The directions matched the predictions written in the test header. Leg A also turned the identity-map case red, which the header did not name; that case reads the mapped table too. Lint, narrowed: eslint --no-inline-config --format json on the 3 changed .ts files gave 3 files, 0 errors, 0 warnings. ① The population is read from eslint.config.mjs (packages/**/*.{ts,tsx,mts,cts}). ② The file count is from the JSON (3). ③ Invariance: the effective parserOptions {ecmaVersion:latest, sourceType:module} have no project, and all 6 active rules are per-file AST rules, so type-aware linting is off and this diff cannot move a verdict on any untouched file. The full pnpm lint is CI's.",
  "gates": {
    "measured_head": "37414c5e8",
    "derivation": "node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack -> 63 commands (change set: 4 paths vs merge base 40b315b03)",
    "run": "63 of 63, each exit code captured to a file before any pipe",
    "first_pass_exit_3": ["pnpm check:dual-build-cjs-loads", "pnpm check:lean-entry-closure", "pnpm check:type-check-debt"],
    "remedy": "turbo run build --filter='./packages/*' --filter='./packages/*/*' (71 tasks), then re-ran all three: exit 0. Also re-ran check:dts-closure (71 packages swept), check:sourcemap-no-sources-content (68) and check:published-files on the full build: exit 0.",
    "final": "63 exit 0",
    "ran_verdict": "dispatch-gates --ran: 63 derived famil(ies) accounted for — 63 run, 0 NOT-MEASURED (a DERIVED zero — all 63 recorded an exit code and none of them is 3)",
    "driver_conformance": "before (789b2ae54): 'OK — 50 covered cell(s), 0 in the DEBT ledger, 0 exempt'; after (37414c5e8): 'OK — 50 covered cell(s), 0 in the DEBT ledger, 0 exempt'",
    "not_measured": "CI-only lanes named by the derivation outside its 63 commands (path-scheduled Test Core / Temporal Conformance / Dogfood / Build Core jobs, the four type-check lanes, workflow-valued families): declared to CI",
    "ci": "in_progress (not awaited)"
  },
  "line_budget": "792 changed lines (+738 / -54) over 4 files vs the 5000 human-merge threshold: under. No governed surface, and no skills/** or ledger line budget in play.",
  "files_changed": [
    ".changeset/20107-turso-remote-external-remote-name.md (+22 / -0)",
    "packages/drivers/driver-turso/src/remote-transport.ts (+96 / -34)",
    "packages/drivers/driver-turso/src/turso-driver.ts (+201 / -20)",
    "packages/drivers/driver-turso/src/turso-local-remote-external-object-parity.test.ts (+419 / -0)"
  ],
  "deviations": [
    "Bounded in-place fix, declared in the PR body: the NOT_IMPLEMENTED/501 refusal of a renaming external.columnMap on the remote data doors. All four conditions hold (same external-binding defect class, the file's own pinned 501 shape, no other claim on these lines, same package tests). It is owed because the resolve alone would have turned a loud failure into a silent []. It is inside the claimed file surface (the remote arms of turso-driver.ts). The remote reclaimSpace route, the constructor and driver-sql's mapping are untouched.",
    "origin/main moved to 50e273fd7 after the merge commit 37414c5e8 (one commit, packages/services/service-analytics only, disjoint from this change set). It was not re-merged. The gates and tests are quoted at 37414c5e8. The remote reclaimSpace card (#20106) had not landed at either reading.",
    "Commit trailers use AGENTS.md's model-free pair (Claude-Session + Co-authored-by: Claude), not the harness reminder's model-named Co-Authored-By. The PR body ends in AGENTS.md's session-URL footer, not the harness reminder's PR ending. AGENTS.md and os-dev.md take precedence, and the pre-push hook refuses a model identifier.",
    "PR labels documentation, size/l, tests and tooling were already present at label-write step ① and were not written by this run. This run wrote only the assignee (os-warren)."
  ],
  "mcp_calls": "0 — no MCP GitHub tool was called",
  "api_writes": "3 — all through the fleet relay (scripts/pm/fleet-write/dispatch.mjs, route dispatch, session_01N8TPEsoJxPsdSdNKGnNGEN), each one POST /repos/objectstack-ai/objectstack/dispatches executed by the fleet-write workflow: (1) pr_create, POST /repos/objectstack-ai/objectstack/pulls, draft #20422 (run 36408832667); (2) label-write --assign os-warren, POST /repos/objectstack-ai/objectstack/issues/20422/assignees (run 36408936784); (3) post-stamped, POST /repos/objectstack-ai/objectstack/issues/20107/comments, this report. No issue_patch was spent: the PR body read back byte-intact. Plus git push of the branch, which is not a REST write.",
  "open_questions": [
    {
      "question": "RemoteTransport is exported from @objectstack/driver-turso, and its 14 data methods gained an optional last 'table' parameter. Nothing was removed or narrowed. Does that additive parameter change the claim's 'Clause-②: no', which the PR body copies verbatim?",
      "options": [
        "A keep 'Clause-②: no' + patch: the parameter is optional, only TursoDriver's own delegate uses it, and RemoteTransport's doc says it is not exposed directly to users",
        "B re-declare as a widening (minor): an exported class's methods accept a new argument"
      ],
      "recommendation": "A. The accepted set of every existing call is unchanged, and the class is documented as an internal delegate. The seat owns the line, so it is flagged rather than changed."
    }
  ],
  "out_of_scope_findings": [
    "class: a · reach: named producer — REST GET /api/v1/data/:object (ObjectQL find -> TursoDriver.find, remote face) on a declared field whose remote column is absent, and service-analytics' ObjectQL aggregate path (AnalyticsService -> engine.aggregate -> TursoDriver.aggregate, remote face) · evidence: at 37414c5e8, over one libSQL file with the local driver as control: aggregate on a mapped table that is really absent gives local DATABASE_ERROR/500, remote []; aggregate grouped by a declared field whose column is absent gives local INVALID_FIELD/400, remote []; find whose WHERE names a declared field whose column is absent gives local INVALID_FILTER/400, remote []. Sources: the RemoteTransport.aggregate catch ('no such table' / 'no such column' -> []) and the RemoteTransport.find $select backstop (returns [] when no projection is present). It applies to managed objects too, and predates this PR · dedupe words: turso remote aggregate no such table empty · turso remote find no such column empty result · RemoteTransport invented empty read · turso remote read exit swallows missing column",
    "carrier: 承接者:无 · noted, not filed — translating external.columnMap on the Turso remote face (it is refused NOT_IMPLEMENTED/501 by this PR; implementing it would put a second copy of the local column rule into the transport compiler, and no demand is measured).",
    "carrier: 承接者:无 · noted, not filed — a remoteName containing a dot: knex splits it into a schema qualifier on the local face, while the remote face quotes it as one name. Unmeasured, and nobody is known to author that shape."
  ]
}

---

Generated by Claude Code