Add protected_branch_ids to gitlab_project_approval_rule

[sirlatrom]

Carry #540 due to fork from wrong repo (just for my own sake).

Fixes #492.

[armsnyder]

I see this is still a draft, but I don't think this will relate to #492, which is about adding a protected_branch_ids attribute to the project_level_mr_approvals resource.

[sirlatrom]

I see this is still a draft, but I don't think this will relate to #492, which is about adding a protected_branch_ids attribute to the project_level_mr_approvals resource.

gitlab_project_level_mr_approvals does not configure protected branches, but gitlab_project_approval_rule does.

In order to properly provide the protected branch IDs, you would either have to configure them using a gitlab_branch_protection or, as will be added in this PR, look up the existing protected branches through a data source.

The current implementation of gitlab_branch_protection unconditionally overwrites any existing protection settings for the chosen branch, which is another issue of concern.

---

Edit: In fact, let me split out the protected_branch_ids to a separate PR.

[sirlatrom]

@armsnyder This PR is now strictly about adding protected_branch_ids attribute to the gitlab_project_approval_rule resource.

[Shocktrooper]

@nicholasklick I saw you were looking over pull requests in this project yesterday. This merge request is also available to be reviewed/merged

[mattkasa]

@sirlatrom this looks great, thank you for adding this! 👍

[mattkasa]

@sirlatrom I'm seeing some acceptance test failures for this:

=== RUN   TestAccGitLabProjectApprovalRule_basic
    testing.go:705: Step 0 error: errors during plan:
        
        Error: "branch": required field is not set
        
          on /tmp/tf-test680098153/main.tf line 53:
          (source code not available)
        
        
--- FAIL: TestAccGitLabProjectApprovalRule_basic (0.19s)
=== RUN   TestAccGitLabProjectApprovalRule_import
    testing.go:705: Step 0 error: errors during plan:
        
        Error: "branch": required field is not set
        
          on /tmp/tf-test574411139/main.tf line 53:
          (source code not available)
        
        
--- FAIL: TestAccGitLabProjectApprovalRule_import (0.14s)

Any chance I could bug you for some help with these?

Seems like an issue with branch = gitlab_project.foo.default_branch. I'm trying to figure out what the right fix is 🤔

[mattkasa]

Sorry for the noise, just talking out loud where I'm at looking at this...looks like gitlab_project doesn't expose a value for default_branch until after apply, and if I add default_branch = "main" to gitlab_project.foo and change gitlab_branch_protection.default branch to branch = "main", then I run into:

=== RUN   TestAccGitLabProjectApprovalRule_basic
    testing.go:705: Step 0 error: errors during apply:
        
        Error: Incorrect attribute value type
        
          on /var/folders/nv/n1lf4tdx0fs9hkl3r88vt0_c0000gn/T/tf-test699686302/main.tf line 105:
          (source code not available)
            |----------------
            | gitlab_branch_protection.default.id is "41:main"
        
        Inappropriate value for attribute "protected_branch_ids": a number is
        required.
        
    testing.go:766: Error destroying resource! WARNING: Dangling resources
        may exist. The full state and error is shown below.
        
        Error: errors during refresh: Incorrect attribute value type: Inappropriate value for attribute "protected_branch_ids": a number is required.
        
        State: <nil>
--- FAIL: TestAccGitLabProjectApprovalRule_basic (2.33s)

Because gitlab_branch_protection.default.id is a two-part ID consisting of project_id and branch.

Not sure where to go next, I thought about adding a branch_protection_id, but that doesn't seem like what you intended either since the test is looking for branch names 🤔

[armsnyder]

+1 to adding a computed branch_protection_id attribute to gitlab_branch_protection, and using it in the test. It could also use a good example in the docs since it's unintuitive.

And because it's related, there's an open PR for fixing the default_branch bug: #583