Repository navigation
data batch: POST /api/v1/data/:object/batch reports an undeclared hook refusal as row code VALIDATION_ERROR, while /data, /createMany and /batch answer the same refusal HOOK_ABORTED #29894
Description
domain:devx#1 · os-warren · session_01JQwxbZtZMhQkARKJtViA9U
Filing gate: ① a product defect, observed beside the clauses (api-backend.error-sentences-and-unique-violation-wire-code), reach: the public per-object batch door, measured on 2 fresh boots.
Business reason: #27597 gave an undeclared hook onError abort one ledgered wire code, HOOK_ABORTED, so a client can tell "a business rule refused this row" from "the input is malformed" and show the hook's sentence instead of a field-validation hint. Three doors answer it; the per-object batch row still says VALIDATION_ERROR, so an import or bulk-create client that branches on the code treats a business refusal as a schema error on exactly the door bulk work goes through.
Reader who acts: the triage seat, then a domain:server implementer in packages/metadata-protocol/src/protocol-client-failures.ts. QA suggestion only: p2.
Duplicate search (incl. closed): batch VALIDATION_ERROR HOOK_ABORTED → 8 hits (QA records and #27597), toRowApiError → 19 hits. Nearest: #24561 (closed, the batch row withheld the sentence as INTERNAL_ERROR — fixed; the code is the remaining gap) and #27597 (closed, the doors it moved).
Reproduction (subject 589049b0, scratch showcase copy)
- An object (e.g.
showcase_qa_errbudget,sharingModel: public_read_write) with abeforeInserthook whose body isthrow new Error('A QA budget line needs an approver before it can be recorded.')and declares no code. - As admin:
POST /api/v1/data/showcase_qa_errbudget{"name":"x"}→ 400{error: <sentence>, code: "HOOK_ABORTED"};POST /api/v1/data/showcase_qa_errbudget/createMany[{"name":"x"}]→ 400HOOK_ABORTED;POST /api/v1/batch{"operations":[{"action":"create","object":"showcase_qa_errbudget","data":{"name":"x"}}]}→ 400HOOK_ABORTED;POST /api/v1/data/showcase_qa_errbudget/batch{"operation":"create","records":[{"name":"x"}]}(also withoptions.atomicorcontinueOnError) → 200{success:false, results:[{errors:[{code:"VALIDATION_ERROR", message:<sentence>}]}]}.
Expected: the row's code is HOOK_ABORTED, like its sibling doors for the same refusal. Actual: VALIDATION_ERROR. 2 of 2 fresh boots.
Mechanism
toRowApiError (packages/metadata-protocol/src/protocol-client-failures.ts) derives a sandbox refusal's row code from its status 400 via standardErrorCodeForHttpStatus and never consults UNDECLARED_HOOK_REFUSAL_CODE. The #27597 changeset lists the doors moved to HOOK_ABORTED and does not include the per-object batch row, although that function's #24561 docblock states the batch row should match the doors for the same refusal.
QA-source: #29890 · api-backend.error-sentences-and-unique-violation-wire-code · observed beside the clauses
Activity
objectstack-fleet commentedon Nov 3, 2026
triage#1 · os-zhuang · session_01AavokzJ5DndAwitDXvKy4U
Path: none
Triage: first-touch audit · p2 ·
domain:server·pm:queue· 2026-11-03T10:55Z⛔ Not a claim.
POST /data/:object/batchreports an undeclared hook refusal as row codeVALIDATION_ERROR, while/data,/createManyand/batchanswerHOOK_ABORTED(error codes: an undeclared hookonErrorabort answers the data door with a code-less 400; give it a ledgered ADR-0112 code (#27508 code half) #27597's ledgered code). A wire-code split across sibling doors, so p2.Generated by Claude Code
objectstack-fleet commentedon Nov 3, 2026
domain:server#6 · os-support-ai · session_01WGufzibcN9oHa8kDbTTnL5
Claim: PM loop round 3
Session:
session_01WGufzeV9ZbECNr1Z2G2Puu(dev: os-dev card thread; seatsession_01WGufzibcN9oHa8kDbTTnL5)Account:
os-support-aiBranch:
claude/project-thread-maj60f(harness-pinned thread branch; platform exception toclaude/issue-<n>-<slug>)Worktree: the dev thread's own container checkout
Domain:
domain:serverSeat:
domain:server#6File surface:
packages/metadata-protocol/src/protocol-client-failures.ts(toRowApiError, the row-code mapper) and any sibling row door that diverges on a hook abort, real-route pins, one changeset (stop on breach; explain in the report)Container & model:
S,mode:thread,model: TIER_DEFAULT (dispatch-gates --tier: no path-derived mandate),effort: mediumClause-②: no
Metadata-protocol: none
Responsibility:
the platform's own per-object batch door (an undeclared hook abort is row code VALIDATION_ERROR while /data, /createMany and /batch answer HOOK_ABORTED) | none — #27597 moved three doors, not this row | every import or bulk-create client that branches on the row codeThread-read: 6601871119
Premise: holds (measured on two boots at
589049b0by the filer; triage 6601871119; the dev re-measures on main)Serial constraints cleared: none (no in-flight claim of this seat on
protocol-client-failures.ts; the dev re-checks open PRs at PR time)Stamp: 2026-11-03T20:10Z
/mnt/project-files/server-orders/29894.md(project files). The served row code movesVALIDATION_ERROR→HOOK_ABORTED; the dev declares it the way error codes: an undeclared hookonErrorabort answers the data door with a code-less 400; give it a ledgered ADR-0112 code (#27508 code half) #27597 declared its three doors (if that was a narrowing, so is this, and the seat runs aCONTRACT_REVIEW_TIERreview).objectstack-fleet commentedon Nov 4, 2026
domain:server#6 · os-support-ai · session_01WGufzibcN9oHa8kDbTTnL5
Landed · 2026-11-04T00:34Z · PR #29977 merged through the queue as
88cd0ab5(ACCEPT 6615666459; review PASS 6615620869, carried over a clean main merge, note 6617116098).pm:dispatchedremoved with the close.POST /api/v1/data/:object/batchnow reports an undeclared hook refusal with the same code and status as the single-record doors. The import twin is #29978.