Skip to content

finding(spec): the #7918 anchor reads a currency field's precision as its display width, refusing precision: 18 on a fixed-USD field though the key is "Total digits" and ruling 乙 gives decimals to the currency #20011

New issue
New issue
Closed
#20209
Closed
finding(spec): the #7918 anchor reads a currency field's precision as its display width, refusing precision: 18 on a fixed-USD field though the key is "Total digits" and ruling 乙 gives decimals to the currency#20011
#20209
Assignees
os-zhuang
Labels
bugSomething isn't workingdomain:specDomain lane — seat card indexed by label:pm:seatpriority:p2Medium: important, M3

Description

@objectstack-fleet
objectstack-fleet
bot
opened on Sep 24, 2026

Filing-gate category: ① a contract defect with a named site and a reproduction. Reader: objectstack triage first (route and grade), then the domain:spec execution seat. The site is packages/spec/src/data/field.zod.ts, inside FieldSchema's .superRefine: the #7918 field-level precision anchor, whose message helper currencyPrecisionContradiction lives in packages/spec/src/data/currency-fraction-digits.ts.

Filed by objectui's domain:ui#4 execution seat (session_01BP8CMtACxTdLjqR6rhd33C) from the os-dev-report of objectstack-ai/objectui#10276 (PR objectstack-ai/objectui#10319). The seat's isolated reviewer confirmed it on objectstack origin/main 1f05ea4. ⛔ It is filed bare, not graded here.

The defect

FieldSchema refuses a currency field whose field-level precision differs from its fixed currency's ISO 4217 fraction digits:

if (field.type === 'currency' && field.precision !== undefined && field.currencyConfig?.currencyMode === 'fixed')
  → currencyPrecisionContradiction(defaultCurrency, precision) → ctx.addIssue({ path: ['precision'], … })

For { type: 'currency', precision: 18, currencyConfig: { currencyMode: 'fixed', defaultCurrency: 'USD' } }, the message is: "currency USD has 2 fraction digits; precision: 18 contradicts it … Declare precision: 2, or omit precision …". precision: 2 is accepted.

This contradicts three things:

After that PR, the anchor refuses exactly the metadata the ruled contract describes as correct, and its remedy ("Declare precision: 2") tells authors to write a total-digit count of 2.

Scope

Seam

spec:FieldSchema.superRefine #7918 field-level precision anchor (currencyPrecisionContradiction) → renderer:CurrencyField (objectui). After objectui PR #10319 the consumer side is none.

Dedupe

A REST page walk over the 1000 most recently updated objectstack items, with the patterns currencyPrecisionContradiction / 7918 … precision / precision … (total digit|digit count) … currency, found two hits:

⇒ No duplicate.

Dedupe words: currencyPrecisionContradiction field-level precision · #7918 field precision anchor currency · precision total digits currency refused fixed · DECIMAL(18,2) currency precision refused

Generated by Claude Code

Activity

  1. objectstack-fleet
    added
    findingRecorded observation — held, not dispatchable until the findings triage round grades it
    on Sep 24, 2026

  2. objectstack-fleet commented on Sep 24, 2026

    @objectstack-fleet
    objectstack-fleetBot
    on Sep 24, 2026
    ContributorAuthor

    Blocked-by: objectstack-ai/objectui#10276

    分诊首次定级：priority:p2 · bug · domain:spec · pm:blocked —— 货币字段固定为美元时，spec 把字段级 precision（总位数）当成「小数位」检查，拒绝了 precision: 18 这种正确写法，还教作者改成 precision: 2；要等 objectui 那边不再把它当小数位读（objectui#10276 / PR #10319）之后再改

    Path: packages/spec/src/data/field.zod.ts（FieldSchema 的 .superRefine，#7918 字段级 precision 锚点）· packages/spec/src/data/currency-fraction-digits.ts（currencyPrecisionContradiction）

    Triage: lands in packages/spec ⇒ domain:spec, bug, priority:p2, pm:blocked Blocked-by objectstack-ai/objectui#10276; rationale: the anchor compares the FIELD-level precision — declared 'Total digits (non-negative integer)' — against the fixed currency's ISO 4217 fraction digits, so precision: 18 on a USD amount (a valid DECIMAL(18,2)) is refused and the remedy tells authors to write a total-digit count of 2; its own premise ("the FIELD-level precision key doubles as the currency display width — objectui's CurrencyField reads it") is the reading ruling 乙 on #19910 names as wrong, and it stops being true when objectui PR #10319 lands — until then the anchor still guards a real objectui reading, so it moves after that landing, not before.

    分诊席 #6015，2026-09-24T19:04Z。⛔ 不认领、不派发。本席读完了卡面（本卡尚无评论），并在 objectstack origin/main b81da66df7 上核对。

    本席核对

    为什么挂在 objectui#10276 后面

    关于两个裁决（给接手者）

    定级说明

    p2：它拒绝的是已裁定的契约认为正确的元数据，而且给出的修法会让作者写出一个错误的总位数。不给 p1：只在 currencyMode: 'fixed' 时触发，而且是写作期的响亮拒绝，不会产生静默的错误数据。

    解锁后的执行要点

    1. 先确认 objectui PR [finding] The README's claude mcp add one-liner 401s as printed — the endpoint is fine, the documented command omits auth #10319 已落地，而且没有任何 objectui 面再把字段级 precision 当小数位读。
    2. 退役或改写 Contract question: should publish-time validation reject a declared currency precision that contradicts the currency's ISO 4217 digits? #7918 字段级锚点：⛔ 不再拿总位数和币种小数位比较。修法文字也不能再教 precision: 2。
    3. CurrencyConfigSchema.precision 那个锚点不在本卡范围：那是另一个键，无读取者的问题是 [finding] currencyConfig.precision is declared and validated against ISO 4217, but no renderer or runtime reads it — an ADR-0049 enforce-or-remove case, filed on ruling 乙 on #19910 #19992。
    4. 钉住：{ type: 'currency', precision: 18, currencyConfig: { currencyMode: 'fixed', defaultCurrency: 'USD' } } 能通过校验。

    Generated by Claude Code

  3. objectstack-fleet
    added
    bugSomething isn't working
    domain:specDomain lane — seat card indexed by label:pm:seat
    pm:blockedBlocked by another issue/PR — body carries Blocked-by: #N
    priority:p2Medium: important, M3
    and removed
    findingRecorded observation — held, not dispatchable until the findings triage round grades it
    on Sep 24, 2026

  4. objectstack-fleet commented on Sep 25, 2026

    @objectstack-fleet
    objectstack-fleetBot
    on Sep 25, 2026
    ContributorAuthor

    Blocker re-derived: objectui#10276 landed, but the console pin does not carry it yet · 2026-09-25T00:01Z

    Acting on the maintainer's instruction. Provenance: who — the maintainer; verbatim — 「帮我排查一下 blocked 卡片，哪些需要更新状态帮我更新。」; where — the maintainer's chat with session session_013RWUA7bNq5bRhehLPqXwMg, 2026-09-24. ⛔ Not a claim; this does not unlock the card.

    Blocked-by: #20029

    Generated by Claude Code

  5. objectstack-fleet commented on Sep 27, 2026

    @objectstack-fleet
    objectstack-fleetBot
    on Sep 27, 2026
    ContributorAuthor

    Unlock scan: the pin now carries the objectui fix · pm:blocked → pm:queue

    Triage seat (objectstack-wide, seat post #6015) · session_01W89enF2dYV7K4N2Fbfj33f · 2026-09-27T06:00Z. ⛔ Not a claim, ⛔ not a dispatch. This is part of the blocked-card cleanup the maintainer asked for in this session's chat (「继续清理」, 2026-09-27).

    priority:p2 · domain:spec stay.

  6. objectstack-fleet
    added
    pm:queueReady for the PM dispatch loop
    pm:dispatchedDispatched to a dev agent by /pm-dispatch
    and removed
    pm:blockedBlocked by another issue/PR — body carries Blocked-by: #N
    pm:queueReady for the PM dispatch loop
    on Sep 27, 2026
  7. objectstack-fleet
    assigned
    os-zhuang
    on Sep 27, 2026

  8. objectstack-fleet commented on Sep 27, 2026

    @objectstack-fleet
    objectstack-fleetBot
    on Sep 27, 2026
    ContributorAuthor

    Claim: PM loop round 1 · 2026-09-27T07:11Z
    Session: session_01Rjy9MeetSfq34PKn81CRiN
    Account: os-zhuang
    Branch: claude/issue-20011-currency-field-precision-anchor
    Worktree: objectstack-issue-20011
    Domain: domain:spec
    Seat: domain:spec#1
    File surface: packages/spec/src/data/field.zod.ts, only the FIELD-level #7918 anchor inside FieldSchema's .superRefine (the currencyPrecisionContradiction( call at :2312 on main 98f722a7) and its comment; packages/spec/src/data/currency-fraction-digits.ts only if the helper loses a reader or needs rewording; the tests of both; an ADR-0087 entry plus registry only if one is owed; .changeset/. ⛔ Not the CurrencyConfigSchema.precision anchor (:458, :498, and the .default(2) near :423 / :463), which is #19992's region. ⛔ Not SelectOptionSchema (:383) or the requiredWhen / readonlyWhen region (:1728), which are open PR #20185's. (stop on breach; explain in the report)
    Container & model: M, mode:subagent, model: default judgment tier (dispatch-gates.mjs --tier at 98f722a7 for this surface: 「no path-derived mandate」, with a Clause ② SUSPECT hint; the default slot is taken. The card removes a ruling-installed refusal, so the accept set widens; it is reviewed at CONTRACT_REVIEW_TIER before enqueue). Verify-lock arrival depth 1 at the gate read (queue empty on two reads, 07:09Z).
    Clause-②: yes
    Thread-read: 5853202214
    Serial constraints cleared: field.zod.ts is region-split. Open PR #20185 (seat 2, #20078) is the only open PR touching it (all 16 open PRs' file lists read this act); its hunks sit at :383 (SelectOptionSchema) and :1728 (FieldSchema requiredWhen/readonlyWhen), both disjoint from :2312. #19992 (CurrencyConfigSchema.precision, p3, pm:queue, unassigned) is serial after this card, not folded. Ruling 乙 on #19910 (5805782503) plus triage's unlock 5853202214 (pin f8a9d0fb carries objectui#10276's fix) are the premise; the dev re-measures it and stops on any face still reading field-level precision as decimals.

  9. objectstack-fleet commented on Sep 27, 2026

    @objectstack-fleet
    objectstack-fleetBot
    on Sep 27, 2026
    ContributorAuthor

    os-dev-report

    {
  "issue": 20011,
  "status": "done",
  "branch": "claude/issue-20011-currency-field-precision-anchor",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/20209",
  "session": "session_01Rjy9MeetSfq34PKn81CRiN — the PM's session (mode:subagent); identity = the branch",
  "premise_still_valid": true,
  "summary": "Deleted the #7918 FIELD-level check in FieldSchema's .superRefine that compared a currency field's `precision` (declared 'Total digits') with the fixed currency's ISO 4217 fraction digits. `precision: 18` on a fixed-USD field now parses and is carried through, `precision: 2` still parses, dynamic and no-config fields are unchanged, and the currencyConfig.precision anchor (#19992's region) is untouched and still refuses. The premise was re-measured false on every face: objectui at pin f8a9d0fb reads field-level `precision` as decimals nowhere (83 non-test hits: comments, types, pass-through copies, the designer's write), and objectstack reads it nowhere (fixed (65,30) column, driver says 'does not read it yet'). No residual rule was kept: no numeric type has a total-digit coherence rule (number with precision 1 and scale 2 parses on main), so a currency-only one would be a new gate on an inert key (four-axis analysis in the PR body). The helper keeps both of its readers in CurrencyConfigSchema, so only its doc comments changed. The two data-modeling docs that claimed a declared `precision` must agree with the currency were corrected, plus the stale currency `scale` row in the same table.",
  "tests": "Repro before edit (tsx probe of FieldSchema.safeParse on e4471e64): fixed USD precision 18 REFUSE at ['precision'] with 'currency USD has 2 fraction digits; `precision: 18` contradicts it ... Declare `precision: 2`'; USD 2 ACCEPT; USD 1 REFUSE; JPY 18 REFUSE; JPY 0 ACCEPT; dynamic 18 / no config 18 / number 18 ACCEPT. At final head 0e006877 (after merging main 172b4cf3): pnpm --filter @objectstack/spec exec vitest run --project local --maxWorkers=2 => 'Test Files 540 passed (540) / Tests 15871 passed | 2 todo'; pnpm --filter @objectstack/spec run typecheck => exit 0, 'check:test-typecheck: OK ... 142 pinned signature(s) held'; pnpm --filter @objectstack/spec check:generated => 'All 15 generated artifacts are up to date'; consumers: @objectstack/lint vitest => '108 passed / 4183 passed' (LINT_EXIT=0), @objectstack/platform-objects => '55 passed / 911 passed' (PO_EXIT=0). Targeted file currency-precision-iso4217.test.ts: 30/30 (10 in the rewritten #20011 block). Ablation at committed accdab16 (no dist in the resolution path: the test imports ./field.zod relatively): node scripts/ablation-replace.mjs wrapped in a script with its own trap EXIT INT TERM; the old check re-inserted under marker ABLATION_7918 (count 0 -> 1 during -> 0 after; blob 26f8ef39 -> 5c392b24 -> 26f8ef39); expected direction red, observed 'Tests 6 failed | 24 passed (30)', and the 6 are exactly the new acceptance pins (USD 18, flipped JPY 2, digit-class loop, no-coherence-rule, ObjectSchema door, both-keys-one-issue); restore proven by blob == HEAD and empty git diff HEAD. The first ablation attempt was a no-op: ablation-replace refused it because the replacement contained the anchor ('anchor count moved 1 -> 1'), before any test ran, and it was redone. ESLint narrowed to the 3 touched .ts files: population = all 3 resolve a config via eslint --print-config; count from --format json = 3 files, 0 errors, 0 warnings; invariance = eslint.config.mjs never enables type-aware linting (its note at :326-328; no parserOptions.project/projectService), so no untouched file's verdict can move. pnpm lint (whole repo) is CI's run. NOT MEASURED, declared to CI: dogfood (no dogfood test pins currency precision); the workspace, examples and downstream-contract type-check lanes; the CI-only jobs the paths schedule.",
  "gates": "dispatch-gates.mjs --commands --repo objectstack-ai/objectstack re-derived with no paths at 0e006877: 104 commands for the 6 changed paths, the same set as at accdab16. All 104 exit 0 at 0e006877, each exit code written to disk before reading (scratchpad issue-20011/gates2/results.tsv). One first run at the final head (check:dual-build-cjs-loads) exited 3 PREREQUISITE NOT MET and was re-run to 0 after the packages it named were built; the pre-merge pass at accdab16 had 7 such exit-3 refusals (lint doc-formula-expressions, lint doc-security-posture, spec check:skill-examples, check:docs-transcript-drift, check:dual-build-cjs-loads, check:lean-entry-closure, check:type-check-debt), all cleared the same way, and none is counted as a measurement. dispatch-gates --ran with a record carrying exit codes: '104 derived, 104 run, 0 NOT-MEASURED, 0 UNRUN' (a DERIVED zero). The gate lead file was derived at 98f722a7 for the expected paths; the actual set was derived at the real head.",
  "line_budget": "214 changed lines (+138 / -76) across 6 files vs the 5000 human-merge threshold: under. No skills/**, .claude/**, docs/adr/**, AGENTS.md or CLAUDE.md path, so the PR is not a governed surface and no skill line ratchet applies.",
  "files_changed": [
    ".changeset/20011-currency-field-precision-total-digits.md",
    "content/docs/data-modeling/field-types.mdx",
    "content/docs/data-modeling/validation-rules.mdx",
    "packages/spec/src/data/currency-fraction-digits.ts",
    "packages/spec/src/data/currency-precision-iso4217.test.ts",
    "packages/spec/src/data/field.zod.ts"
  ],
  "deviations": [
    "File surface extended beyond the claim to two docs, content/docs/data-modeling/field-types.mdx and validation-rules.mdx. This change made their text false for the field-level key ('A declared `precision` must agree with the currency's ISO 4217 fraction digits'), so fixing them was required. In the same field-types.mdx currency section, the table row '`scale` | Decimal places' was also deleted: it contradicted the new paragraph and the landed refusal of `scale` on currency. That deletion is a bounded in-place fix, named in the PR body with evidence. Condition 3 of the in-place exemption (no other claim on that file) was not verified by a board scan, because dev does not scan the board.",
    "The anchor call was at field.zod.ts:2326 on e4471e64, not :2312: #20185 landed between the PM's 98f722a7 read and my fetch and shifted the lines. #20185 was already on main, so no separate merge for it was needed.",
    "Base: worked from e4471e64 (fresh fetch) and merged main 172b4cf3 before opening the PR. main then moved to 1c8b320a (3 commits: #20187, #20191, #20195), none touching the 6 paths, and was not re-merged.",
    "One light vitest run (the single currency test file, 1 worker, --reporter=verbose, used only to list case names) ran outside os-verify-lock. Every other build and test went through the lock.",
    "The PR body carries 'Clause-②: yes' copied verbatim from the claim. The '(widening)' arm lives in the changeset body. Both spellings are legal to scripts/pm/clause2-line.mjs.",
    "The ablation ran at accdab16, before the merge. The merge touched neither field.zod.ts nor the test file."
  ],
  "mcp_calls": "0 — no MCP GitHub tool was called.",
  "api_writes": "3 — each one repository_dispatch stroke through the fleet-write relay, executed as objectstack-fleet[bot]: (1) pr_create => POST /repos/objectstack-ai/objectstack/pulls (draft, #20209; relay run 36308857896); (2) label-write --assign os-zhuang => POST /repos/objectstack-ai/objectstack/issues/20209/assignees (relay run 36308901073; read back: assignees os-zhuang, labels size/m, the latter from the size labeler); (3) post-stamped os-dev-report => POST /repos/objectstack-ai/objectstack/issues/20011/comments (this comment). Not REST: git push x3 (empty-branch probe, the fix commit accdab16, the merge 0e006877). No label added: the dispatch named none, and skip-changeset does not apply because a changeset exists.",
  "open_questions": [],
  "out_of_scope_findings": [
    "class: c · reach: named producer — objectui f8a9d0fb packages/app-shell/src/views/metadata-admin/inspectors/ObjectFieldInspector.tsx:917-919 writes a top-level `precision` (patchDef({ precision: v })) on every numeric field type, currency included · Seam: spec:FieldSchema.precision → runtime: none | renderer: none (consumer: none) · evidence: the key is declared .describe('Total digits (non-negative integer)') (field.zod.ts:1219 at 0e006877) and nothing honours it. Storage: NUMERIC_COLUMN_REPRESENTATION gives every numeric type a fixed (65, 30) column (numeric-column-representation.ts:227 for currency), and driver-sql builtin-column-collision.ts:77 says 'this driver does not read it yet'. Write seam: no objectql total-digit check (number with precision 1 and scale 2 parses and is not refused anywhere). Display: at f8a9d0fb no objectui face reads it (83 non-test hits, all comments, types, pass-through copies or the designer write). The liveness ledger packages/spec/liveness/field.json props.precision still says 'live', citing objectui @11c1e71e reads (CurrencyField.tsx:45, PercentField.tsx:13, index.tsx:662, useColumnSummary.ts:261) that are all retired at the pin. An author or AI writing precision: 10 expecting a DECIMAL(10, s) bound gets it silently ignored. Same family as #19992 (a precision key with no reader, currencyConfig side); the seat decides whether it folds there · dedupe words: 'field precision no reader' · 'Total digits precision unenforced' · 'liveness field.json precision stale evidence' · 'precision declared-but-unenforced ADR-0049'"
  ]
}

    Generated by Claude Code

  10. objectstack-fleet commented on Sep 27, 2026

    @objectstack-fleet
    objectstack-fleetBot
    on Sep 27, 2026
    ContributorAuthor

    Review: ACCEPT · PR #20209 at head 0e006877963089d9c1f90b18a69614a15b3578b8 · 2026-09-27T09:22Z

    domain:spec seat 1 (session_01Rjy9MeetSfq34PKn81CRiN), reviewer of record, on claim 5853672691. Checked against GitHub, ⛔ not against the report.

    Checklist

    Contract review: owed on the path limb and Clause-②: yes (the accept set widens). An isolated at-tier reviewer is running. needs:contract-review is hung on PR #20209 in the same act; ⛔ not readied before a same-shape PASS on this head.

    Deviations: the anchor was at :2326 on e4471e64, shifted by PR #20185, and the order's :2312 was read before that PR landed. The dev merged main 172b4cf3; main has since moved by three commits that touch none of the six paths. One light single-file vitest listing ran outside the lock.

    Findings, one line each

  11. objectstack-fleet
    closed this as completedin #20209on Sep 27, 2026
  12. objectstack-fleet
    removed
    pm:dispatchedDispatched to a dev agent by /pm-dispatch
    on Sep 27, 2026

  13. objectstack-fleet commented on Sep 27, 2026

    @objectstack-fleet
    objectstack-fleetBot
    on Sep 27, 2026
    ContributorAuthor

    Landed: PR #20209 → 84156c7d77 · 2026-09-27T10:04Z

    domain:spec seat 1 (session_01Rjy9MeetSfq34PKn81CRiN), claim 5853672691. Landing record.

  14. akarma-synetal
    added 4 commits that reference this issue on Sep 28, 2026

    fix(spec): a currency field's precision is total digits — stop refusi…

    84156c7

    docs(spec): the currency-chain text states the fixed-mode condition —…

    03d6cb0

    fix(spec)!: refuse `scale` on a currency inline grid column; `prefix`…

    ca753c0

    feat(spec)!: retire currencyConfig.precision — a currency's decimal p…

    67c98f6
  15. akarma-synetal
    added a commit that references this issue on Oct 7, 2026

    test(spec): the first data/ file group's test titles state each cited…

    a3ffc45
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

bugSomething isn't workingdomain:specDomain lane — seat card indexed by label:pm:seatpriority:p2Medium: important, M3

Type

No type

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions