Evaluating size labels on PRs

[bsmth]

Hi everyone 👋🏻 there's been a few requests for the team over the last few months to add size labels to pull requests, mainly for two reasons:

• help estimate the complexity of a PR before reviewers start (of which size is one factor)
• look over historical data or PRs over time to see the distribution of contribution size

We already have logic for labeling PRs by technology category based on file paths, so I have rolled it into the existing labeler workflow. Hopefully we will merge this shortly and see how it works for reviewers and contributors.

Update (Jan 30th '24):

Going with the following labels for mdn/content for now:

Size	LoC
XS	0-5
S	6-50
M	51-500
L	501-1000
XL	1001+

• Update PR: chore(ci): Increase PR size labels from S to XL content#31986

Update: The labeler landed here:

• chore(ci): Switch out labeler action, define labels in workflow inline instead of by config file content#31758

Criteria for size labels:

I did not put too much thought into the size criteria for now, which initially looks like this:

          # Configuration for 'extra small' PRs
          xs_label: "size/xs"
          xs_max_size: "5"

          # Configuration for 'small' PRs
          s_label: "size/s"
          s_max_size: "10"

          
          # Configuration for 'medium' PRs
          m_label: "size/m"
          m_max_size: "50"

          # Configuration for 'large' PRs
          l_label: "size/l"
          l_max_size: "100"

          # 'Extra large' PRs are anything over `l_max_size` (>100 LoC)
          xl_label: "size/xl"

Ignoring files:

I have set the ignore list to the following:

          # List of files to exclude from size calculation
          # Files matching these patterns will not be considered when calculating PR size
          files_to_ignore: |
            "files/en-us/_redirects.txt"
            "files/en-us/_wikihistory.json"

If anyone has opinions or feedback on this, you're welcome to share. If you think other files should be ignored, the format of labels (size/xs, size/s, size/m, size/l, size/xl) or the label diff size is not good for each label (FYI, see how k8s do it), feel free to share.

Thanks a lot!

[Rumyra]

Thanks for opening the discussion @bsmth !

Just want to mention we're trialling this as a possible help with pull requests and the process - it was something we had feedback on when we we're discussing large prs https://github.com/orgs/mdn/discussions/501 (not explicitly in that thread but iirc in the November community meeting) 👍

[bsmth]

After some testing, we ended up swapping out for a different GH action and we've landed the labeler in this PR:

• chore(ci): Switch out labeler action, define labels in workflow inline instead of by config file content#31758

You can see it in action here:

mdn/content#31684 (commits)

If there's any issues or feedback, feel free to let us know here or through the usual channels (GH issue, chat, etc.)

Thank you!