DUE TO SPAM, SIGN-UP IS DISABLED. Goto Selfserve wiki signup and request an account.
News, Updates, Topics of Interest
| Query open no reviews | https://github.com/apache/flink/pulls?q=+state%3Aopen+type%3Apr+review%3Anone+draft%3Afalse | 374 | ||||||
| Query open | https://github.com/apache/flink/pulls?q=+state%3Aopen+type%3Apr++draft%3Afalse+ | 421 | ||||||
| Query non stale open using label | https://github.com/apache/flink/pulls?q=+state%3Aopen+type%3Apr++draft%3Afalse+-label%3Astale+ | 178 | ||||||
The 21st 30 minute working meeting of the Community Health Initiative (CHI) for Eastern time zone.
- 14 PRs came in this week
- Thomas Cooper is creating statistics from the Flink git repo and is close to getting out graphs and statistics to evidence what we might want to improve.
- David Radley amended Flip-518 for the community review process in line with community feedback to move it to a git action out for vote.
David Radley produced a dev list post for the Kryo v2 to v5 migration.
PRs to merge this week
https://github.com/apache/flink/pull/26481
https://github.com/apache/flink/pull/26478
PRs to merge from previous week
https://github.com/apache/flink/pull/26449 - trivial change
https://github.com/apache/flink/pull/26462 - trivial change
https://github.com/apache/flink/pull/26450 - looks safe and fixes a security issue
Pull requests list
[FLINK-37608][fs-connector] File sink supports to close buckets in parallel when snapshot component=Connectors/FileSystem#26498
[FLINK-37693][state] Avoid reading NOT_OWNED file by local file system when restoring component=Runtime/StateBackends#26497
[FLINK-37711] Remove unused getJobGraph from PipelineExecutorUtils component=Client/JobSubmission#26493
[FLINK-36902] Remove support for FlinkSecurityManager component=<none>#26491
[hotfix] Rename SessionDispatcherFactory to DefaultDispatcherFactory#26490
[FLINK-37700][table-planner] Fix SQL parse exception when creating a view with PTF component=TableSQL/Planner#26487
[hotfix][core] Optimize the merge logic lines calling in Histogram.java.#26486
[hotfix][core] Replace the concating on strings with the placeholder in log lines#26485
[hotfix][core] Minor polish for CliFrontend.java#26484
[FLINK-37308] Support pauseOrResumeSplits in HybridSource component=Connectors/Common#26483
[FLINK-37698] Rename fromSchedulerNg to fromWithinExecutionGraph component=Runtime/Task#26482
[hotfix][test] Remove redundant keywords and lines for ForStGeneralMultiGetOperationTest.java#26481
[hotfix][runtime] Delete the unused BoundedInputStream.java.#26478
[FLINK-36767] Bumped cyclonedx-maven-plugin to 2.9.1 to resolve CVE-2024-38374#26476
Ideas for "Flink bot" implemented in a git actions
- automatically run spotless apply and make those changes to the code.
- check whether code in java main code has code in java test
- identify API / config modules that might require documentation changes.
- Police Jira titles to make sure they match the process
- Make sure the Jira referred to the in the title exists and has a descriptions
Previous meeting actions
- Thomas Cooper raised a PR around pre commit hooks linting, we discussed and agreed to look at a consistant approach with the server and client side linting pre commits.
- Thomas Cooper to propose the stale bot on other Flink repos starting with the operator Kafka then JDBC connectors .
- We talked about the community review and agreed that I will raise a Flip for it, I was hoping for more support on the dev list but we think a Flip is an easier thing to ask for support for.
- We talked about the v1 -> v2 state as documented in the v2 release notes I reviewed this week. The feedback is that the advice on state migration will come after v2 is released. The worry is this is a an issue all existing Flink users will hit and we do not have any guidance.
- Pedro is up for doing community reviews and may help us look into the state migration.
- David Radley started discussion on dev list for community review with the idea to raise a Flip to change the process and update the flink bot to enable this process.
- We talked about a contributor role being able to assign Jiras to themselves. But are aware that the current process Flink Jira Process#Assignee seems to indicate that a committer would assign a Jira if the committer was going to merge the PR. It would be nice to have the work assigned to individuals so it is visiblely owned.
- Thomas Cooper mentioned that he had started a piece of work to move the overnight checks onto the PR checking. Including spotless apply. The idea is to move the checks into Git Actions.
- We talked about the ongoing change of Kyro 2 to 5 and how this impacts V1 to V2 migration.
- David Radley to ask the community for Tom and I to have the authority to add labels - this would allow us to fix up the PRs and add the stale label to them
- when we have this we talked of using ready-to-merge label, and reviewed-by-CHI
- this could be done in a contributor role that we had previously talked about with Robert Metzger
- Thomas Cooper to look into using flinkbot on Apache infrastructure
- We talked about the v1 → v2 state discussions and the breaking change - David Radley to review
- Thomas Cooper to suggest on dev list reducing the times to stale and time to close and change so it checks less (we setting it checking a lot of PRs very quickly so we could get the stale label on everything quickly)
- Thomas Cooper to look into the CI / flink bot - we talked of moving this to be a Git action
- labelling whether the CI has failed for open PRs would be helpful; to reduce the number that need reviewing to those that are mergable.
- David Radley to analyse what we have in the non draft , open unstale PR bucket so we can see how they group and look at actions based on the data.
- Thomas Cooper to look into the CI / flink bot - we talked of moving this to be a Git action
- We talked of moving the Flink / CI bot type activity to a Git action. Robert Metzger said there was some work in this area that had not yet completed. Thomas Cooper and I offered to help complete this
- We talked of creating a list of things that the Flink bot should do, I have created a starter list, please let us know if there are other ideas you want to add.
- We talked about amending the Flink operator release process to include creating OLM Operatorhub versions. There are instructions in IBM that we are looking to open source and share to automate. We will track this in this meeting.
Older PRs - reviewed in the past - mostly waiting for CI to pass to review again.
TODO check in meeting
- 25592: CI Failed, they need to run spotless to fix. Reviewed already by David Radley, author has not responded.
- 25590: CI Passed, simple docs update Left comment Needs Committer review
- 25583: CI Failed, duplicate of 22660 which is related to FLIP-317 and could be closed.
- 25576 CI Passed. This is a trivial change adding additional debug information. Can be merged.
- 25575 CI Passed. There is little to no description of what this PR is about and the JIRAs are empty? Asked for more context to help review.
- 25540 CI Failed, still working on fix for scala tests.
- 25538 CI Failed, still working on fixed for CI failures.
- 25528 CI Failed, still working on fixes.
- 25524 CI Failed, still working on fixes