From 814c1a9d05905b688808c6f0a43457ba74ab6889 Mon Sep 17 00:00:00 2001 From: Claude Date: Sun, 6 Sep 2026 01:40:56 +0000 Subject: [PATCH] docs(examples): cite the #6318 assertion by content, not by line address (objectui#7853) The citing docblock addressed `safe-validate-corpus-6318.test.ts` by line number. A comment-only edit to that file -- every diff line a comment, both revisions transpiling byte-identically -- moved the cited assertion from line 98 to line 117, making the address false. Nothing in the repository went red. Cite the assertion's content and name the enclosing `it(...)` title instead, and drop the line number. A quoted assertion rots too if someone rewrites the assertion, but it rots LOUDLY: the quote stops matching anything, and one `git grep -F` says so. The quote is the shortest span that is unique tree-wide (1 occurrence); the shorter `'type' in o` matches 2 sites, and the longer form only adds an `expect(...).toBe(false)` wrapper that can be restyled without changing the claim. Co-Authored-By: Claude Opus 5 Claude-Session: https://claude.ai/code/session_01KbJQ1y1J12nZxYzFWhP8Q3 --- .../component-fixture-declared-keys.test.ts | 33 ++++++++++++++----- 1 file changed, 24 insertions(+), 9 deletions(-) diff --git a/examples/schema-catalog/test/component-fixture-declared-keys.test.ts b/examples/schema-catalog/test/component-fixture-declared-keys.test.ts index f89b3cb30f..931a192052 100644 --- a/examples/schema-catalog/test/component-fixture-declared-keys.test.ts +++ b/examples/schema-catalog/test/component-fixture-declared-keys.test.ts @@ -832,15 +832,30 @@ describe('catalog corpus: no overlay-menu item authors the undeclared `value` ke * * ## What already covered this, and why it is not the PAIR * - * `safe-validate-corpus-6318.test.ts:98` asserts `options.some((o) => 'type' - * in o) === false` — but against the STATIC IMPORT of `basic-select.json` - * alone. That is 3 of the 66 select options in the corpus: exactly the one - * fixture #6902 repaired. A second select authoring `type`, in any of the - * other six categories, is invisible to it. That is the objectui#7072 failure - * mode named on this same file — a pin that covers part of a class while - * READING as though it covers the class. This block is the pair-scoped half; - * the #6318 pin keeps its own narrower job (that fixture's three options carry - * `label`) and is deliberately left untouched. + * `safe-validate-corpus-6318.test.ts` asserts `options.some((o) => 'type' in + * o)` is false — in the `it` titled *basic-select: the third option carries + * `label`, which is the only child SelectItem renders* — but against the + * STATIC IMPORT of `basic-select.json` alone. That is 3 of the 66 select + * options in the corpus: exactly the one fixture #6902 repaired. A second + * select authoring `type`, in any of the other six categories, is invisible to + * it. That is the objectui#7072 failure mode named on this same file — a pin + * that covers part of a class while READING as though it covers the class. + * This block is the pair-scoped half; the #6318 pin keeps its own narrower job + * (that fixture's three options carry `label`) and is deliberately left + * untouched. + * + * ⚠️ That citation names the assertion's CONTENT and carries NO line number, + * deliberately (objectui#7853). It used to carry one, and a comment-only edit + * to that file — every diff line a comment, both revisions transpiling + * byte-identically — moved the assertion from line 98 to line 117. An edit + * that changed no behaviour whatsoever falsified this sentence, and not one + * instrument in this repository went red. A quoted assertion rots too, if + * someone rewrites the assertion, but it rots LOUDLY: the quote stops matching + * anything and a single `git grep -F` says so. ⛔ Do not put the line number + * back. The quote is the SHORTEST span that is unique tree-wide (1 occurrence; + * `'type' in o` alone matches 2, the other being this file's own sweep below), + * and it deliberately stops short of the `expect(...).toBe(false)` wrapper, + * which can be restyled without changing the claim this paragraph leans on. * * ## Why a structural sweep and ⛔ not `.success` *