Repository navigation
spec: the padded grouping-field refusal is trim-equivalence under UNICODE whitespace, but its .describe()/docs/ADR entry all say just "whitespace" #17962
Description
PR #17498 narrowed GroupingFieldSchema.field to refuse leading/trailing whitespace. The prose that ships with it — the .describe(), the generated doc page, and the ADR-0087 semantic entry — all say "whitespace", but the pattern's accept set is JavaScript's \S, which refuses considerably more than the space and tab a reader will picture.
Filed by the domain:spec execution seat, session session_01MkQhmuuJAVDjmeWNixwDDH, 2026-09-13T08:24Z, out of the at-tier post-hoc review of PR #17498. ⛔ No domain:* or priority:* applied — routing and grading are triage's. ⛔ Not a defect: the code is right, the description of it is incomplete.
The measurement
The at-tier round proved trim-equivalence by exhaustion — 14,641 four-character strings over an 11-symbol alphabet including U+00A0 (NBSP), U+3000 (ideographic space) and U+FEFF (BOM): 0 divergences from s === s.trim(). Lit control on the same harness (/^\S*$/): 150 divergences, so the harness discriminates.
⇒ the accept set is exactly trim-equivalence, Unicode included. A name with a leading NBSP or a trailing ideographic space is refused, by a rule whose published description mentions neither.
Why this is the failure mode worth fixing
These are precisely the characters an author does not know are in their string. A name pasted from a spreadsheet, a CJK editor, or a BOM-prefixed CSV export carries them invisibly. The author reads a refusal that says "whitespace", looks at a name with no visible space in it, and has no way to reconcile the two. The diagnostic already names the offending spelling — but a spelling whose difference is invisible is not a name the author can act on.
What is owed
Name the class in the three prose sites (.describe() on GroupingFieldSchema.field, the regenerated content/docs/references/ui/view.mdx row, and the ADR-0087 entry 18.ui-list-view-grouping-field-padded-refused): the rule is trim-equivalence under Unicode whitespace, not ASCII space. Worth considering alongside it: rendering the offending characters in the diagnostic as escapes ( ) rather than as themselves, so the message shows the author what they cannot see.
Previously raised as "Advisory 2" by an earlier review round on that PR and left unaddressed — filing it so it has a carrier this time rather than a mention.
Generated by Claude Code
Activity
objectstack-fleet commentedon Sep 24, 2026
Closed
not_plannedby the director seat (summon #28 续,session_01GLdRPcbaCBQCTvVmU6YEUY), 2026-09-24T03:36Z, on the maintainer's word — closure review batch 1 of the opendomain:speccards under the restructured triage standard, presented card by card from the business angle in this seat's chat; maintainer verbatim: 「16524 改文档,不需要按组织取;17493 回收;其他同意」.Four cards close in this group, each with the reason the maintainer approved:
I18nLabelSchemaform 1), so no capability is missing — only the inline-map shortcut, which no product row on NORTH-STAR asks for. The schema refusal is loud and correct (优先级 4 satisfied); the linter crash was fixed by PR #16280. Re-file only when a customer asks for inline locale maps on objects or fields.spec-changes.jsonand its five Zod schemas ship with the package; only the JSON-Schema face is absent, and the twogen:docswarnings are pinned byCATEGORIES_WITHOUT_SCHEMA_CLOSURE(triage: 「不会静默腐烂」). No external consumer of the file is named ⇒ 优先级 2 「不在清单上 ⇒ 不做」. Re-file when a consumer that needs the JSON-Schema face is named.sharing_rule.condition): 「仪器为车队服务 … 说不出就不做」 — no fleet decision or product behaviour changes when the binding lands. Parent #18589 is closedcompleted; PR #18797 already records why the entry is unbound.Ledger: seat post #12708, this summon's next block. ⛔ No repository file edited by this act.