philo-he commented on issue #12288: URL: https://github.com/apache/gluten/issues/12288#issuecomment-5125913300
> Adding one recent fan-out snapshot (not aggregate data): x86 run [30218534234](https://github.com/apache/gluten/actions/runs/30218534234) expanded to 41 jobs / 1,595.8 runner-minutes (929.3 Spark, 604.1 TPC). Separately, [#12630](https://github.com/apache/gluten/pull/12630) changed one line under `ep/build-velox/**` and triggered x86, ARM, and Enhanced—48 workflow jobs in total. > > Would a report-only changed-component classifier be useful first—only recording the proposed CI tier without changing coverage—so we can collect evidence before deciding whether to split a linked sub-issue? @Smallfu666, Thanks for the comment. The current workflow is a bit conservative. That said, we do have cases where a small change in a build script can affect a Spark test, for example, the build options for simdjson can change JSON parsing behavior. I'm not sure what the best practice is for CI trigger control. We may have to make a tradeoff: avoid wasting too much GHA resource, while relying on the daily full test run to catch any problematic merged PR. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
