Skip to content

Commit 9a41a11

Browse files
authored
docs: request changes for correctness and performance findings in the review skill (apache#6483)
main needs a single approval to merge, so a correctness or performance finding posted as a comment does not stop another committer's approval, or an armed auto-merge, from merging the PR over it. A Request changes review from a committer does, until the same reviewer approves or the review is dismissed. Define the findings that must block the merge, add a Review State item to the output format, cover the re-review that lifts the block, and say how to post the review when the user asks for it.
1 parent 11a27c3 commit 9a41a11

1 file changed

Lines changed: 30 additions & 0 deletions

File tree

‎.ai/skills/review-comet-pr/SKILL.md‎

Lines changed: 30 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -166,6 +166,25 @@ before merge. If it is, raise it and expect a response. If it is not, cut it. Th
166166
- This bar is about what gets raised, not about how it is worded. Keep the tone below. A question you
167167
expect an answer to still counts as something the author needs to address.
168168

169+
## Request Changes for Correctness and Performance Regressions
170+
171+
Two kinds of finding have to block the merge until the author addresses them: a correctness problem
172+
that the PR introduces, and a performance regression. A correctness problem means Comet can return a
173+
different answer from Spark, including a result where Spark raises an error, or `Compatible()` over a
174+
known divergence. Crashes, hangs, leaks, and lost or corrupted data count too. A performance
175+
regression means an existing path gets slower or uses more memory. A tracking issue for a later fix
176+
does not address either one.
177+
178+
When any finding is one of these, the review is submitted as **Request changes**. `main` needs a
179+
single approval to merge, and any committer's approval counts, including one given before this
180+
review. A **Comment** review does not stop that approval from merging the PR over the finding. A
181+
**Request changes** review from a reviewer with write access blocks the merge until the same reviewer
182+
approves the PR or someone with write access dismisses the review.
183+
184+
Requesting changes commits the reviewer to the re-review. Once the author has addressed the findings
185+
behind it, the reviewer approves the PR or dismisses the review. A later **Comment** review leaves
186+
the block in place.
187+
169188
## Output Format
170189

171190
Present your review as guidance for the reviewer:
@@ -177,6 +196,12 @@ Present your review as guidance for the reviewer:
177196
5. **Suggested Review Comments**, specific comments the reviewer could leave, with file and line
178197
references. Everything here is something you expect the author to address. Anything that did not
179198
clear the bar above should not appear.
199+
6. **Review State**, how to submit the review. Use the first case that applies:
200+
- **Request changes** when any finding is a correctness problem that the PR introduces or a
201+
performance regression. Name those findings.
202+
- **Approve**, or dismiss the earlier review, when this is a re-review and the findings behind the
203+
reviewer's earlier **Request changes** review have all been addressed.
204+
- **Comment** otherwise.
180205

181206
## Review Tone and Style
182207

@@ -203,3 +228,8 @@ Instead:
203228
**IMPORTANT: Never post comments or reviews on the PR directly.** This skill and all of its siblings
204229
are for providing guidance to a human reviewer. Present all findings and suggested comments to the
205230
user. The user will decide what to post.
231+
232+
When the user tells you to post the review for them, submit it in the review state from the output.
233+
For **Request changes** that is
234+
`gh pr review <pr> --repo apache/datafusion-comet --request-changes --body-file <file>`. A PR comment
235+
or a **Comment** review does not block the merge.

0 commit comments

Comments
 (0)