feat(spec): reject a declared currency precision contradicting the currency's ISO 4217 fraction digits (#7918) - #8232
Merged
Conversation
…rrency's ISO 4217 fraction digits (#7918) Maintainer ruling 2026-08-12, Option A. Two anchors, one shared verdict: the field-level precision key (no default — checked in FieldSchema's superRefine) and CurrencyConfigSchema.precision (default(2) relocated into a post-check .overwrite() so the superRefine sees authored-vs-absent; parse output stays byte-identical). Fires only when the currency is statically known (currencyConfig fixed mode); dynamic mode and non-CLDR codes fail open by design. Digit table is a checked-in CLDR 48.0 snapshot with provenance. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0123k4cam2jEAkPmbJeoaY3r
…rency-precision-iso4217
…ion optionality (#7918) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0123k4cam2jEAkPmbJeoaY3r
…rency-precision-iso4217
|
The latest updates on your projects. Learn more about Vercel for GitHub. 1 Skipped Deployment
|
Contributor
📓 Docs Drift CheckThis PR changes 1 package(s): 106 hand-written doc(s) reference the affected code and may need an implementation-accuracy re-verification:
⛔ 7 release-owned page(s) also reference the affected code. These are read-only:
|
github-actions
Bot
added
size/m
documentation
This was referenced Aug 12, 2026
huangyiirene
marked this pull request as ready for review
August 12, 2026 22:18
huangyiirene
pushed a commit
that referenced
this pull request
Aug 12, 2026
…tree (#7872 relay lap over #8230/#8232) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0123k4cam2jEAkPmbJeoaY3r
huangyiirene
pushed a commit
that referenced
this pull request
Aug 13, 2026
os-regen merge lap for landing: merge origin/main (committed first, per scripts/pm/os-regen-merge.sh), then wholesale regen of the os-regen'd references/api pages from the merged tree. Sibling assertions on the regenerated tree: #8230 retirement state (def names only in retirement prose + migration entry), #8232 currency fraction-digit guard, #8234's 13 comparand-door exports in api-surface/data.json, #8236's groupBy union row all survive; MONGODB_MULTI_TENANT_UNSUPPORTED stays absent from the ledger and the references ErrorCode union reads main-minus-one (+265 -> +264; the union grew by 2 on main since the previous lap). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0123k4cam2jEAkPmbJeoaY3r
This was referenced Aug 13, 2026
akarma-synetal
pushed a commit
to akarma-synetal/framework
that referenced
this pull request
Aug 17, 2026
… enforced once at the shared compile face for all five drivers (objectstack-ai#7872) (objectstack-ai#8234) * feat(spec): comparand-type door — accepted literal comparand set enforced at the shared compile face (objectstack-ai#7872) The measured superset string | number | bigint | boolean | null | Date is defined in packages/spec (filter-comparand-type.ts), enforced by parseFilterAST on everything it returns and by the engine's lowering seam on the object form; everything else is refused with the INVALID_FILTER / 400 envelope naming the accepted set. bigint is narrowed copy-on-write to its exact number (refused beyond 2^53), which kills the driver-memory mingo crash cell without touching the frozen driver; the mongo silent-edit cell ({qty: undefined} wiring to {} = match everything) becomes a refusal at the door. SQL-family allowlists source their set membership and wording from the door instead of local copies. FILTER_COMPARAND_TYPE_CASES conformance table added and consumed by all five driver suites. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0123k4cam2jEAkPmbJeoaY3r * chore(spec): regenerate api-surface + export-origins for the comparand-type door exports (objectstack-ai#7872) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0123k4cam2jEAkPmbJeoaY3r * test(objectql): re-judge the deep-equality operator-map pin under the objectstack-ai#7872 door Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0123k4cam2jEAkPmbJeoaY3r * chore(spec): regenerate api-surface + export-origins from the merged tree (objectstack-ai#7872 relay lap over objectstack-ai#8230/objectstack-ai#8232) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0123k4cam2jEAkPmbJeoaY3r * test(objectql): pay the two farm reds — registerObject arity, and un-erased engine option bags (objectstack-ai#7872) The door suite's registerObject(deal) call was missing the required packageId (the +1 over objectql's frozen TEST_DEBT 355); and its option bags used bare 'as any', which the query-options-erasure ratchet counts (240 -> 245). The bags now type-check uncast — FilterCondition's index signature admits the deliberately off-contract comparands, which is the point of the door — except the one FilterArray case, which carries the sanctioned 'as unknown as EngineQueryOptions' naming the contract it bypasses (not counted). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0123k4cam2jEAkPmbJeoaY3r --------- Co-authored-by: Claude <noreply@anthropic.com>
akarma-synetal
pushed a commit
to akarma-synetal/framework
that referenced
this pull request
Aug 17, 2026
…ger — a boot refusal never reaches a wire envelope (objectstack-ai#8035) (objectstack-ai#8239) * fix(spec): MONGODB_MULTI_TENANT_UNSUPPORTED leaves the error-code ledger (objectstack-ai#8035) The code was registered under @objectstack/driver-mongodb (objectstack-ai#3724) but no response envelope can carry it: the CLI boot handler rethrows the tenancy refusal pre-HTTP and aborts, and the one request-reachable trigger (assertObjectsNotTenantScoped via syncObjectSchema inside the metadata protocol's ensureObjectStorage) is swallowed by a documented best-effort catch. Registered-but-unemittable is the class the ledger header names; precedent OVERLAY_PERSISTENCE_FAILED / objectstack-ai#5783 — this instance is the sister shape (never had a wire path, rather than lost it), now documented in the header. The boot refusal stays untouched: MULTI_TENANT_UNSUPPORTED_CODE, the error class, its message, and the CLI's duck-typed literal match. A new driver-mongodb test pins the code literal (with the ledger row gone it is the only cross-package guard on the literal serve.ts matches); the ledger test now asserts the wire vocabulary refuses the retired string. Reference docs regenerated (ErrorCode enum 264 -> 263). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0123k4cam2jEAkPmbJeoaY3r * docs(spec): regenerate references from the merged tree (objectstack-ai#8035) os-regen merge lap for landing: merge origin/main (committed first, per scripts/pm/os-regen-merge.sh), then wholesale regen of the os-regen'd references/api pages from the merged tree. Sibling assertions on the regenerated tree: objectstack-ai#8230 retirement state (def names only in retirement prose + migration entry), objectstack-ai#8232 currency fraction-digit guard, objectstack-ai#8234's 13 comparand-door exports in api-surface/data.json, objectstack-ai#8236's groupBy union row all survive; MONGODB_MULTI_TENANT_UNSUPPORTED stays absent from the ledger and the references ErrorCode union reads main-minus-one (+265 -> +264; the union grew by 2 on main since the previous lap). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0123k4cam2jEAkPmbJeoaY3r --------- Co-authored-by: Claude <noreply@anthropic.com>
akarma-synetal
pushed a commit
to akarma-synetal/framework
that referenced
this pull request
Aug 17, 2026
…ered/virtualScroll (ADR-0049, objectstack-ai#7176) (objectstack-ai#8252) * refactor(spec)!: retire pass-through-only list-view keys striped/bordered/virtualScroll (ADR-0049, objectstack-ai#7176) Maintainer ruling 2026-08-10: every measured reader copies the key forward and ObjectGrid, where the chains end, never applies any of the three — copy-without-apply is dead in effect. retiredKey() tombstones on ListViewSchema (radiating into ObjectListViewSchema), ADR-0087 D2 conversion view-list-passthrough-keys-removed + D3 chain step + six RETIRED_KEYS_BY_MAJOR[17] entries, ledger rows flipped to dead with the retirement note, form inputs pruned, docs updated, changeset with FROM->TO. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0123k4cam2jEAkPmbJeoaY3r * test(spec): match the striped prescription's actual casing in the pin test Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0123k4cam2jEAkPmbJeoaY3r * merge origin/main (os-regen artifacts taken from main; regeneration follows) * chore(spec): regenerate artifacts after merging origin/main (4dc1c7d relay: objectstack-ai#8230/objectstack-ai#8232/objectstack-ai#8234/objectstack-ai#8236/objectstack-ai#8239) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0123k4cam2jEAkPmbJeoaY3r --------- Co-authored-by: Claude <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes #7918
Implements the maintainer's Option-A ruling (issue comment, 2026-08-12, provenance 「7918 A,7917 ②,7900 收敛两扇门,7929 来源标记」): publish-time validation rejects a declared currency
precisionthat contradicts the currency's ISO 4217 / CLDR fraction digits, when the currency is statically known.What changed
Two anchors, one shared verdict (
currencyPrecisionContradictionin the newpackages/spec/src/data/currency-fraction-digits.ts):CurrencyConfigSchema.precision— checked in asuperRefinebefore the default materializes. The ruling's design trap governs the shape: the property carried.default(2), which bakes in at parse, so authored-vs-defaulted is indistinguishable post-parse. Measured anchor choice: the.default(2)moved off the property into a post-check.overwrite(), so the refinement seesundefinedfor an unauthored precision (authored-only firing), and the parse OUTPUT stays byte-identical to the.default(2)era — measured on origin/main (37b82ed) before the change and pinned in tests:{"precision":2,"currencyMode":"fixed","defaultCurrency":"JPY"}for an untouched fixed-JPY config, same shape order, same materialized default..overwrite()rather than.transform()per the measured finding:FormViewSchema.groupsis declared as "alias to sections" but nothing folds it — every consumer reads onlysections#6926 precedent inview.zod.ts(a pipe has no.extendand breaks shape introspection); its one accepted cost, same as there:CurrencyConfigParsednow infersprecision?at the type level while the runtime value is always a number (ADR-0122 forbids hand-narrowing the alias).precision— checked inFieldSchema's existingsuperRefinechain (the same anchoring as the siblingstorage.notNull × requiredWhenrule, so every publish door inherits it). This key has no schema default, so authored-vs-absent needed no relocation.Statically known means
currencyConfig.currencyMode: 'fixed'. One premise refinement against the ruling's parenthetical "(field-levelcurrency, …)": the spec's strictFieldSchemadeclares no field-levelcurrencykey — that spelling is an objectui-side read of looser column configs, and writing it on a spec field is already an unrecognized-key rejection (filed the missing-guidance gap as #8163). The only statically-known-currency form the spec can express is fixed-modecurrencyConfig, and both precision keys are judged against it. In authored-fixedmode the check also fires whendefaultCurrencyitself was defaulted (CNY): the fixed currency is statically known either way, and the precision was authored — pinned in tests.Deliberately partial, by design (do not "improve"):
dynamiccurrencyMode has no single currency to check against and is untouched; codes outside CLDRcurrencyData(crypto/custom, e.g. the existing BTC-precision-8 test) fail open — refusing unknown codes is a different rule nobody ruled.Error message names both numbers, per the ruling: "currency JPY has 0 fraction digits;
precision: 2contradicts it — …", plus the prescription (declare the currency's own count, or omitprecision).Alias funnel:
CurrencyConfigSchema'sdecimals/scalealiases are strict-rejection-with-suggestion (not renames), so an alias spelling cannot carry a contradicting width past the check — the author is pointed atprecision, where the check waits. Pinned.The digit table
Static, checked-in snapshot of CLDR 48.0
currencyData(ICU 78.2, node v22.22.2 full-icu), 162 codes, generated locale-free viaIntl.NumberFormat(...).resolvedOptions().maximumFractionDigits— provenance and regeneration snippet in the module header. Checked-in rather than probed at validation time so the verdict is deterministic (a small-icu host answers 2 for everything) and the validation path takes no Intl dependency. Renderers keep deriving from live Intl (objectui#4361); both read CLDR, so they agree. The card's measured anchors (0: JPY/KRW/CLP/ISK/VND; 2: USD/EUR/CNY/GBP; 3: KWD/BHD/OMR/TND) are pinned against the table.In-repo census (PM assumption #3)
Three
currencyConfigdeclarations repo-wide (examples/app-showcase: field-zoo, semantic-zoo, account), allfixed+USD+precision: 2— agreeing. Zero contradictions to fix; no field-levelprecisionon any in-repo currency field. Census clean, as expected.Bump: minor, per the #3746 precedent
Acceptance narrowing in the #3746 shape (strict-closing family, shipped as Minor Changes in the 17.0.0-rc line): every newly-rejected input is a contradiction no consumer could honor — renderers already derive the width from the currency when precision is absent, and the rejected combinations rendered money with digits the currency does not have. Not major (
check:changeset-no-majoralso refuses majors); the changeset argues the same and records the one type-level visible change.Verification
@objectstack/spec: fullpnpm test— 386 files / 10216 tests passed;pnpm typecheckgreen;check:generated— all 13 artifacts up to date after regenerating the one it proved stale (content/docs/references/**, a 1-line optionality rendering forcurrencyConfig.precision; rides in this PR). Re-verified after the finalorigin/mainmerge (fa48973 side: speccontracts/*moved — rebuilt, all 13 still green).custom, pathprecision/currencyConfig.precision, message naming both digit counts); USD+2, JPY+0, KWD+3, absent-precision, dynamic-mode (authored and defaulted), unknown-code (BTC+8) accepted — the agreeing combos byte-identically (stringified parse outputs measured on origin/main 37b82ed before the change, pinned as literals).const n: number = cfg.precisionagainst the rebuilt dist.d.ts→ red (TS2322) exactly as predicted,number | undefinedcontrol clean — proves the declaration change is real and rebuilt, not cached. B (anchor choice): temporarily restoring.default(2)on the property predicted 3 reds; observed 4 — the three predicted (noisy-shape guard, its byte-identity row, the field-level agreeing-acceptance test) plus the field-level JPY+2 rejection test, which failed on issue ORDER: the config-level check over-fires on the baked default and itscurrencyConfig.precisionissue displaces the field-level issue as first. One more red than predicted, same mechanism — the baked-default over-firing shape the relocation exists to prevent, proven live from the committed state, then restored viagit checkout.@objectstack/metadata-protocol75 files / 1094 tests passed;@objectstack/platform-objects16 files / 347 tests passed (both after building their dependency closures — the first lap's exit-1 was the fresh-worktree unbuilt-closure false red, not a regression);objectstack validategreen in all three example apps (showcase / crm / todo).Docs:
content/docs/data-modeling/field-types.mdxcurrency section states the rule, its partiality, and the authored-only firing.Generated by Claude Code