Configure merge queue for the main branch

[mnonnenmacher]

Please enable the merge queue as part of the branch protection rules for the main branch as this can currently not be done via Otterdog (see eclipse-csi/otterdog#86).

The required settings are:
Merge method: "Rebase and merge"
Maximum pull requests to build: 5
Minimum pull requests to merge: 1 or after 5 minutes
Maximum pull requests to merge: 5
Only merge non-failing pull requests: true
Consider check failed after: 60 minutes

[eclipse-apoapsis-bot]

settings have been applied manually to the existing branch protection rule for branch main:

[mnonnenmacher]

Thanks, looking good in my test PR eclipse-apoapsis/ort-server#1.

[mnonnenmacher]

@netomi There is one issue with the merge queue that is caused by the global eclipsefdn/eca workflow. This workflow is required but not triggered by the merge queue, therefore it times out after a while (see eclipse-apoapsis/ort-server#2):

github-merge-queue bot removed this pull request from the merge queue due to no response for status checks

To solve this the merge_group event would have to be added as a trigger to the workflow:

on:
  merge_group:
    types: [checks_requested]

What would be the correct place to request such a change?

[netomi]

So the eclipsefdn/eca status check is not handled via a workflow but through a GitHub App that reacts on webhook events.
We will need to request that here: https://gitlab.eclipse.org/eclipsefdn/it/api/git-eca-rest-api

As a short term solution, we can remove the eclipsefdn/eca check from the list of required checks in the branch protection rule.
Change the line https://github.com/eclipse-apoapsis/.eclipsefdn/blob/main/otterdog/eclipse-apoapsis.jsonnet#L22

to

-          required_status_checks+: [
+          required_status_checks: [

so the eca check is not inherited from the default config.

[netomi]

Created ticket: https://gitlab.eclipse.org/eclipsefdn/it/api/git-eca-rest-api/-/issues/161

[mnonnenmacher]

Thanks for the explanation and creating the ticket. I will apply the workaround for now.