Handle Rate Limit

[pattacini]

Instead of letting the action fail because of having hit the rate limit, we may consider handling it through suitable wait states by reading relevant info via the API.

See:

• https://octokit.github.io/octokit.rb/Octokit/Client/RateLimit.html
• https://octokit.github.io/octokit.rb/Octokit/RateLimit.html

[pattacini]

Done in #24 ✔

[JeroenKnoops]

@pattacini FYI

Although this solution works, I have changed the workflow a little bit, and instead of using a PAT, we're now using a github app to do all the API calls.

We've used something similar to: https://github.com/peter-murray/workflow-application-token-action

You can grant the exact permissions to an app, and with the workflow-application-token-action you can get a temporary token to do all API calls. This will no longer case any rate limit to be hit since the number is set a lot higher for apps.

See: https://docs.github.com/en/developers/apps/rate-limits-for-github-apps

Workflow will look something like:

      - uses: peter-murray/workflow-application-token-action@v1
        name: Retrieve App Token
        id: app_token
        with:
          application_id: ${{ secrets.APP_ID }}
          application_private_key: ${{ secrets.APP_PRIVATE_KEY }}

      - name: Outside collaborators
        run: |
          cd scripts
          ./outside-collaborators-handler.rb
        env:
          OUTSIDE_COLLABORATORS_GITHUB_ORG: ${{ github.repository_owner }}
          OUTSIDE_COLLABORATORS_GITHUB_TOKEN: $ {{ steps.app_token.outputs.token }}

I can't really make a PR, because without creating an app it's hard to tell if it works in your case.

[pattacini]

Thanks a lot @JeroenKnoops for this nice suggestion!
I may give it a try in the upcoming days. Either way, I'll be documenting somehow this addition from the main README.md

[pattacini]

Hi @JeroenKnoops

I've incorporated your suggestion in the main README (see 46d1ee1), but with a second thought, I believe this further layer would add an unnecessary complication now as we rarely hit the rate limit.

I'll certainly be considering this again in the future if I'll start out experiencing downgraded performances in our context.

Thanks heaps for the nice feedback!

[JeroenKnoops]

We were having a lot of rate limit hits, because we were using the same user also for other actions like gathering metric with CHAOSS. That's why this was our solution. I just wanted to share it for the future or when other people have similar issues! :)