Controlling Build Scope for Dependent Projects in Pull Request Builds

[ivandjordjevic]

We are using a single repository with multiple AL projects and have alwaysBuildAllProjects set to false. For example:

1. Project A (base project)
2. Project B (depends on Project A)

When we make a change to Project B, only Project B is built during the Pull Request Build, which works as expected.

However, when we change Project A, both A and B are built, even though Project B wasn’t modified directly.

Question:
Is there a way to configure AL-Go so that a change in Project A only triggers a build for A, and not for B, unless B is also explicitly modified?

[freddydk]

Today, that is only possible by splitting the apps into 2 repositories
How else would you determine that a change in app a doesn't break app b?

[ivandjordjevic]

Yes, I understand the need for that but I'm trying to bypass it.
Problem is Project B has lots of Tests and it takes hours to execute. So it would also help if tests could run only in the modified projects if we can't skip the build.

[freddydk]

Why would you have to have these apps in the same repository?

[freddydk]

The current design of AL-Go is that every CI/CD build is tested and can be released. By running a mode, where apps are built if they are modified and not if dependencies have changed, would cause the latest build to be "maybe" OK.
You would basically have to run a full build before releasing - which is against the current design.
Recommendation for a scenario like yours would be to split the repo in two and use GitHub packages for dependency resolution.

[ivandjordjevic]

Thanks, that makes sense. I was just wondering since we can already skip tests with doNotRunTests, maybe there was way to apply it conditionally in PR builds to avoid running tests or even building projects that weren’t modified.

We have 11 apps in the repo, all released together, so we rely on the scheduled Test Current workflow to run everything and ensure the branch is always in a releasable state.

[freddydk]

You can definitely set doNotRunTests on some workflows and not others.
One way would be to just set "doNotRunTests" to true in global repo settings and then for Test Current, Test Next Minor and Major - in the workflow specific settings - set it to false - then you will be running tests there.
You can also set it for CI/CD workflow if you like.