Enforce commits are following our conventions via github actions

[raghunathd8]

Adds a commitlint check job that ensures the following rules:

1. conventional commits
2. supported commit types: feat, fix, dep, doc, ci, refactor, test
3. Breaking changes can be indicated with a ! before :, e.g. feat!: new breaking feature.
   Breaking changes must have a breaking-change: or breaking change: trailer. This is case in-sensitive.

The commitlint configuration ensures those settings. The GitHub Actions job executes them and fails the PR validation if any of the commits in a PR are not compliant. Those can then be amended as needed by the PR author.

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: peanball / name: Alexander Lais (0e532d3)
• ✅ login: raghunathd8 / name: Raghunath Deshpande (d16d02a, d16d02a)
• ✅ login: Soha-Albaghdady / name: Soha Alboghdady (d16d02a)

[peanball]

The trailer is in the wrong format in the commit, should be:

Co-authored-by: Soha Alboghdady <soha.alboghdady@sap.com>

< vs. (.

[maxmoehl]

@plowin if you want to run the CI you have to approve the PR ;)

[peanball]

@Soha-Albaghdady it seems that GitHub actions is picky about what it supports as base image on public runners. Right now all PRs are blocked because none of the commitlint steps run.

https://docs.github.com/en/actions/using-github-hosted-runners/about-github-hosted-runners#supported-runners-and-hardware-resources

See also https://stackoverflow.com/questions/70959954/error-waiting-for-a-runner-to-pick-up-this-job-using-github-actions

So maybe we need to change back to ubuntu-latest and install npm.

There is an action for that though, that allows you to select the NPM version. Action resources should also be cached and should be the preferred way over downloading via script on the fly.

https://github.com/actions/setup-node

[raghunathd8]

https://docs.github.com/en/actions/using-github-hosted-runners/about-github-hosted-runners#supported-runners-and-hardware-resources

This is exactly what I was referring to, github have selected specific runners that they support and encourage to use.

[peanball]

This is exactly what I was referring to, github have selected specific runners that they support and encourage to use.

It seems that they just don't support others than the listed ones.

[peanball]

one final bit for the commit, as this is a change on the overall CI setup, I would propose the prefix ci: .

[peanball]

I think the checkout with depth 0 may be counteracting the commit lint checking of a commit range. Could you please clarify?

[peanball]

Checked with "breaking change" indicator ! without the appropriate header in action build https://github.com/cloudfoundry/pcap-release/actions/runs/5545026407/jobs/10123338288?pr=76

works.

changes applied and ok. I applied further changes that I don't want to approve myself.

[a18e]

see comments/suggestions, otherwise great!

[b1tamara]

LGTM

[a18e]

If you omit the space before the subject (e.g. dep(scope)!:Subjectstarts), commitlint will fail with the error type may not be empty.
I find this confusing/misleading.
The following suggestions fix this but in a slightly hacky way (also the rule name breaking-change-indicator doesn't fit anymore).
I'm fine with the suggestion being ignored and approve of the rest of the PR.

[peanball]

If you omit the space before the subject (e.g. dep(scope)!:Subjectstarts), commitlint will fail with the error type may not be empty.
I find this confusing/misleading.
The following suggestions fix this but in a slightly hacky way (also the rule name breaking-change-indicator doesn't fit anymore).
I'm fine with the suggestion being ignored and approve of the rest of the PR.

As mentioned on the other comment, it's too hacky for me. Also for the purpose of our initial goal, i.e. "don't allow misfit commits", this does not add anything.

If you wanted, you can propose a new checker (via separate PR) that does more than type-empty from the default set and inspects the commit message and all its elements in full. But I see this separate to the breaking change indicator, and a non-vital improvement.