ci: automate APK size comparison

[mixelas]

Purpose / Description

Automate APK size comparisons for pull requests by comparing the PR release APK against the latest successful main release APK.

Fixes

• Fixes APK size optimization umbrella (remaining task: size comparison on PRs) #16626

Approach

The emulator workflow records and uploads the release APK size as an artifact for PRs and main pushes. A follow-up workflow downloads the PR artifact and the latest successful main artifact, compares their sizes, and updates a single PR comment with the result.

How Has This Been Tested?

Verified by GitHub Actions: unit tests, Kotlin/JavaScript quality checks, and release emulator tests passed.

Checklist

Please, go through these checks before submitting the PR.

• You have a descriptive commit message with a short title (first line, max 50 chars).
• You have commented your code, particularly in hard-to-understand areas
• You have performed a self-review of your own code
• UI changes: include screenshots of all affected screens (in particular showing any new or changed strings)
• UI Changes: You have tested your change using the Google Accessibility Scanner

[david-allison]

Ran git diff --check successfully and reviewed the final diff.

This isn't OK for a CI change

[mixelas]

I’ve corrected the github-script pin and verified the updated workflow with CI. The unit and emulator tests passed. Thanks for pointing out the gap in my validation.

[david-allison]

Verified by GitHub Actions: unit tests, Kotlin/JavaScript quality checks, and release emulator tests passed.

Are you testing the comparison itself?

[mixelas]

I can test the size calculation locally with sample artifact values, but the artifact download and PR comment require GitHub Actions. The full workflow will therefore be verified when the PR-triggered emulator workflow completes and invokes the comparison workflow.

[david-allison]

Can you test it on your fork?

[mixelas]

I tested the comparison math and invalid-size handling locally. I haven’t verified artifact download or PR commenting yet. Since this uses workflow_run, GitHub requires the listener workflow to be on the default branch before it can run. I don’t want to merge it into my fork’s default branch just to test, would you prefer I move the comparison into the emulator workflow so it can be exercised before merge, or is post-merge validation acceptable?

[david-allison]

I typically test on my fork's main, you can force push to origin/main without issue to undo dodgy commits

[mixelas]

I tested the workflow on my fork. The PR emulator run passed, and the comparison workflow successfully downloaded both APK-size artifacts, calculated the difference, and posted a comment. The test showed 49,910,777 bytes for both builds. I restored my fork’s main afterward and closed the temporary test PR.

[david-allison]

Cheers! Could you link the results?

[mixelas]

Sure! The PR emulator run passed here, and the APK size comparison run completed here. It compared both APKs at 49,910,777 bytes and posted the result on the temporary test PR.

[david-allison]

Forgot to submit this

[david-allison]

This seems to be empty?

gh api repos/ankidroid/Anki-Android/actions/runs/36271038668 --jq '.pull_requests'

returns

[]