[ 
https://issues.apache.org/jira/browse/FLINK-40995?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18127183#comment-18127183
 ] 

sepuri sai krishna commented on FLINK-40995:
--------------------------------------------

I have a one line fix for this. Please assign it to me.

> Draft pull requests are not excluded from the community review label scan
> -------------------------------------------------------------------------
>
>                 Key: FLINK-40995
>                 URL: https://issues.apache.org/jira/browse/FLINK-40995
>             Project: Flink
>          Issue Type: Bug
>          Components: Build System / CI
>    Affects Versions: 2.4.0
>            Reporter: sepuri sai krishna
>            Priority: Minor
>
> The community review workflow means to skip draft pull requests, but the 
> filter never
> removes any, so drafts are scanned and can be labelled community-reviewed.
> In .github/workflows/community-review.sh the filter is written with jq's
> update-assignment operator instead of a comparison:
>   select((.node.isDraft = false))
> "=" assigns false to .node.isDraft and returns the modified object, which is 
> always
> truthy, so select keeps every element.
> On a page of 100 open pull requests of which 10 are drafts:
>   select((.node.isDraft = false))   -> 100 kept   (current)
>   select(.node.isDraft == false)    ->  90 kept   (intended)
> The workflow logs show the same. Every page of the run on 2026-10-14 08:06 
> retained
> everything it received:
>   Filtering  100 received pull requests...  100 PR retained
>   Filtering  100 received pull requests...  100 PR retained
>   Filtering  100 received pull requests...  100 PR retained
>   Filtering   96 received pull requests...   96 PR retained
> That is 396 of 396 retained, while 31 of the open pull requests are drafts. 
> Four draft
> pull requests currently carry the community-reviewed label: FLINK PR 26288, 
> 26810,
> 28330 and 28831.
> Fixing the filter stops new mislabels. It does not remove the four existing 
> ones, since
> the script's delete logic only keeps the two review labels mutually exclusive 
> and never
> reconciles against draft status.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to