As part of our triage process, we look at the incoming PRs and try and assign them to the right users.
Recently, the triage group has been under significantly increased pressure and I wanted to measure how much exactly. Sharing the data here with a couple explanatory caveats:
- Is a PR merged or not: is answered via the “Merged” label on github. This can miss some reverted and you will see a blip in 2022 due to our Infra not assigning it properly.
- Is a PR from regular contributors or not: answered via the “open source” label. Which is the one used to non-regular contributor’s PRs. Specifically, all of these are the ones that hit the triage group.
Here is the raw data:
The key takeaway for now is clear:
- Yes we’re getting a lot more PRs than we used to.
- This increase is both from regular contributors and new contributors
- The merge rate is not following the submission rate at all. My conjecture is from two things: a lot of these are low quality PRs that we’re not interested in merging and that the repo was already review bound, so more open PRs doesn’t mean more things are going to be merged.
The two obvious next steps from here are:
- Improve triage process to reduce the burden of filtering out these lower quality PRs
- Improve the review workflow and reviewer base to be able to properly review (and thus merge) more PRs
