Issue: Incorrect history & notification text when a pull request is auto-unapproved

[Alex Ronke]

I have a complaint about how Bitbucket describes automated approval retraction for a changed Pull Request. 

Right now, there isn't a distinction in the PR history or notifications between a reviewing user that deliberately retracts their approval versus when the PR automatically retracts that approval upon the branch receiving a new commit.  Since the reviewer isn't the one actually un-approving the PR, the text as currently shown is incorrect, as it states that the approval retraction was that user's own action.

To recreate:

Make a pull request for a branch with commits.

Add a reviewer.  We'll call her "Barb".

Have "Barb" approve the request.

Push a new commit to the branch.

What currently happens:

I receive a notification that "Barb" has marked the PR as unapproved.

The PR overview/history shows that "Barb" marked the PR as unapproved.

What should happen:

The notification should communicate in some manner that Bitbucket has retracted Barb's approval (and the approval of any other reviewers) because a new commit was added to the branch.

The PR overview/history should also show that Bitbucket retracted all approval(s) because a new commit was added to the branch.

[Julius Davies _bit-booster_com_]

Your question inspired us to implement an improved "auto-unapprover" in our paid add-on, PR-Booster for Bitbucket Server.

Here's the config screen:

And here's the end-result:

Our add-on allows the auto-unapprover to be enabled globally, per-project, or per-repo.  We also implemented functionality for retaining "needs work" reviews, which normally get dropped the moment the PR receives any new commits.

We use a special "PR-Booster - Approval Reset Bot - New Commits Detected" service user to make the unapprove actions understandable on the PR history.

Install PR-Booster for Bitbucket Server to try this out!