Repository navigation
docs(genai): recount the closed structs and cite the grammar where it is stated - #2110
Open
justinchuby wants to merge 1 commit into
Open
justinchuby wants to merge 1 commit into
justinchuby wants to merge 1 commit into
Conversation
… is stated Two facts in section 10.4 and section 10.6 describe the code inaccurately. Both were carried unchanged through 2668208, which corrected the sentences around them; a repair certifies its own neighbourhood, so a stale fact inside one is harder to doubt than a stale fact on its own. The count is stale because this design's own implementation moved it. #2009 added structures to schema/ir.rs, so `grep -c deny_unknown_fields` went 45 to 46 at 47fda6c and reads 49 on main. Both sites said 45. The derivation is now stated beside the number so the next reader can recompute it rather than trust it. The citation resolves but does not say what cites it. `schema/mod.rs:62-78` covers additive fields and the token_packed reshape; the accepted grammar and the normalization rule are at 51-60. Section 10.6 claimed grammar and normalization from 62-78 and now cites 51-60; section 10.4 claimed the whole normalized contract and now cites 51-78, whose head line is the grammar. Verified by content, not by range: each cited span is asserted to contain the symbol or phrase the sentence claims for it, and the count is asserted against the grep it names. capability_catalogue, which include_str!s this document, selects 3 and passes 3. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: Copilot <223556219+Copilot@users.noreply.github.com>
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #2110 +/- ##
==========================================
+ Coverage 80.32% 80.73% +0.41%
==========================================
Files 426 429 +3
Lines 204769 214808 +10039
Branches 204769 214808 +10039
==========================================
+ Hits 164484 173431 +8947
- Misses 34664 35611 +947
- Partials 5621 5766 +145
Flags with carried forward coverage won't be shown. Click here to find out more. 🚀 New features to boost your workflow:
|
This branch has not been deployed
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
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.
Two facts in the metadata decisions document describe the code inaccurately. Both are in passages that
266820801("describe the shipped schema version gate accurately") rewrote, and both were carried through it unchanged.That is the part worth stating plainly: a correction commit is the worst place for a stale fact.
266820801is more accurate than what it replaced, but everything it touched now reads as verified-today, including the two facts it merely carried past. An assertion sitting inside a checked repair is harder to doubt than one nobody claimed to have checked.1. The count is stale, and this design's own implementation moved it
schema/ir.rsno longer carries 45deny_unknown_fieldsattributes.The document said 45 in two places (§10.4 and §10.6). #2009 — the implementation of the very surface these sections describe — added the structures that moved it. The number was true when written and was falsified by the code it documents.
Fixed to 49, and the derivation is now stated beside it (
grep -c deny_unknown_fields) so the next reader can recompute it instead of trusting it. A count in prose has nothing that can falsify it; naming the command it came from gives it one.2. The citation resolves, and does not say what cites it
schema/mod.rs:62-78is a real range inside the right doc comment on the right field — and it covers additive fields and thetoken_packedreshape. The accepted grammar and the normalization rule are at 51-60.mod.rs:62-78mod.rs:51-60mod.rs:62-78mod.rs:51-78§10.4 claims the whole contract, so it now spans 51-78, whose head line is the grammar sentence. §10.6 claims grammar and normalization specifically, so it narrows to 51-60.
This is the failure mode this document argues about, committed by the document: a citation that still resolves is the most durable way to be wrong, because a broken link announces itself and a wrong-but-resolving one does not.
Verification
Content-aware, not line-in-range — each cited span is asserted to contain the symbol or phrase the sentence claims for it:
capability_catalogue, whichinclude_str!s this document, selects 3 (confirmed with--listbefore reading the result) and passes 3. Fullonnx-genai-metadata: 344 passed, 0 failed.git diff --checkclean. Docs-only: one file, four factual corrections and one reflow, no contract or design change.Credit
Both defects were found by the schema agent auditing
266820801against the code it describes, after I had checked onlyversion.rs:42,78by content and left the third citation and the count uninspected.