docs: require audit skill to file issues and add Spark 4.1.1 to version list#4468
Open
andygrove wants to merge 3 commits into
Open
docs: require audit skill to file issues and add Spark 4.1.1 to version list#4468andygrove wants to merge 3 commits into
andygrove wants to merge 3 commits into
Conversation
The audit-comet-expression skill previously permitted leaving semantics-decision findings as prose recommendations in the PR description, on the assumption the reviewer would pick them up. In practice that note dies with the PR. Tighten Step 6 and Step 7 so that every high-priority finding either becomes an inline fix + test, or a filed GitHub issue + ignored regression test, before the audit PR is opened. Add a dedicated 'Findings that need follow-up' subsection spelling out the workflow (search, file, cross-reference from the support-doc sub-bullet and the PR description). Surfaced while re-running the string-expressions audit in apache#4461: several higher-risk findings (CometCaseConversionBase compat gating, StringRepeat negative-count divergence, translate grapheme semantics, bit_length/octet_length BinaryType, decode legacy flags) were left as prose only and had to be filed retroactively as apache#4462-apache#4467.
Spark 4.1.1 is now a tracked release in the project, so the audit skill should pull and diff against it alongside 3.4.3, 3.5.8, and 4.0.1. Update every Step 1 / Step 2 / Step 5 reference and the sub-bullet template at Step 8 accordingly.
This was referenced May 27, 2026
Merged
The string-expressions audit (PR apache#4461) revealed six recurring failure modes where the skill documented findings rather than acting on them. Strengthen the consistency checklist and auto-fix list to close these loopholes: - Add checklist item 10: expression-shape restrictions (literal-only argument, child data type, etc.) must be declared in `getSupportLevel`, not gated inside `convert` with `withInfo`. Cite `CometLeft` / `CometRight` / `CometSubstring` as the canonical example. - Add checklist item 11: Spark 4.0+ collation routing through `CollationSupport.X.exec` and `StringTypeWithCollation` means the expression is `Incompatible` for non-default collations. Link apache#4496 as the umbrella issue and reject "behaviour unchanged for `UTF8_BINARY`" as a justification for `Compatible`. - Add checklist item 12: a sub-bullet that says "Known divergence" or "Known limitation" on a `Compatible` branch is a smell. The skill must promote the support level rather than documenting the divergence in prose only. Cite the `replace` empty-search-string case. - Add checklist item 13: unreachable serde registrations (e.g. the `btrim` mapping for `StringTrimBoth`, which is rewritten by `RuntimeReplaceable` before serde runs) must be deleted, not catalogued. - Add an issue-verification step to the reason-wording guidance and the follow-up-issue workflow. Every cited issue must be opened with `gh issue view` to confirm it exists, is open, and matches the divergence before the URL ships in a reason string or support-doc sub-bullet. Add the matching auto-fix patterns to Step 7's "apply fixes automatically" list so future audits resolve these inline rather than filing them as prose follow-ups.
| --- | ||
| name: audit-comet-expression | ||
| description: Audit an existing Comet expression for correctness and test coverage. Studies the Spark implementation across versions 3.4.3, 3.5.8, and 4.0.1, reviews the Comet and DataFusion implementations, identifies missing test coverage, and offers to implement additional tests. | ||
| description: Audit an existing Comet expression for correctness and test coverage. Studies the Spark implementation across versions 3.4.3, 3.5.8, 4.0.1, and 4.1.1, reviews the Comet and DataFusion implementations, identifies missing test coverage, and offers to implement additional tests. |
Contributor
There was a problem hiding this comment.
I think it is 4.0.2 now?
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.
Which issue does this PR close?
Closes #.
Rationale for this change
Two improvements to the
audit-comet-expressionskill surfaced by the string-expressions audit in #4461.What changes are included in this PR?
correctness/documentationlabel, cross-reference from the support-doc sub-bullet and the PR description).v4.1.1to everyfor tag in ...loop in Step 1 / Step 2, add a4.0.1 → 4.1.1row to the cross-version diff list, and update the descriptions, Step 8 sub-bullet template, and Step 5 output-format section to enumerate four Spark versions.How are these changes tested?
Skill-only documentation change. The next per-category audit will exercise both new behaviours: it will clone Spark 4.1.1 alongside the existing versions, and any high-priority finding it cannot fix inline will be filed as a tracking issue rather than left as prose.