Hey everyone, I’d like to propose a small improvement to how our PR bot assigns reviewers in https://github.com/apache/beam/pull/40315 and gather lazy consensus before merging.
Currently, the PR bot assigns and @-tags one reviewer for every matching label in .github/REVIEWERS.yml. Because component PRs almost always match both a specific component label and a language/infra label (e.g., kafka + Java + Build), the bot frequently tags 2–5+ reviewers on a single PR. This leads to: 1. Unclear ownership: When multiple reviewers are tagged across broad and specific labels, it isn't clear who is primarily responsible for the review. 2. Extra notification noise: General SDK/build reviewers are co-tagged on PRs that already have a more specific component reviewer assigned. To solve this, I created https://github.com/apache/beam/pull/40315. The ;tldr is that it will assign a single reviewer, with more narrow labels (like Iceberg) prioritized over broader ones (like Java). The exact mechanics are described in more depth in the PR, but in my experimental simulations it dropped notifications by ~20% without imbalancing reviews. Please let me know on this thread or on t <https://github.com/apache/beam/pull/40315>he PR if you have any feedback. If there are no objections in the next 72 hours, I'll go ahead and merge this under lazy consensus. Thanks, Danny
