[pr] Investigate is how we would handle the commits and coverage in the case of a fast-forward merge.

[drazisil-codecov]

As far as I can tell, we do not handle this type of merge. If we don't should we? What would be the work involved?

From: https://sentry.zendesk.com/agent/tickets/137884

[eliatcodecov]

It doesn't seem like we handle fast forward merges properly, leading to coverage not being properly moved to the default branch. We have dedicated logic for squash and merge, but not fast forward. So the default behavior in this case is not clear. Investigation ask is to determine how we handle FF merges and then we can decide if we need to fix anything / how to fix.

[giovanni-guidini]

I don't think we need to handle this case explicitly...

Looking at the function that updates the commits we first check if the PR is merged, and then if it was merged via squash. From my understanding this is not the case for a fast forward merge (where the feature branch is just added to the main branch.

If it's not a squash merge (which I think it's the case) we move the commits to the main branch (exactly what we want).

If we're not seeing the commits in the correct branch then that can leave 2 other options:

1. We're not listing the commits on the PR correctly
2. We're not identifying that the PR was not merged via squash correctly

I should probably try to reproduce the error 😅

[giovanni-guidini]

To test this I used this repo: https://github.com/giovanni-guidini/example-python

Main branch commits:

1st test: giovanni-guidini/example-python#1
(merged via "merge"... I expected a FF-merge but there is a merge commit there. Still the commit from the feature branch was moved to main)

Codecov lists that commit in the main branch as expected (and the merge commit afterwards)

The 2nd test: giovanni-guidini/example-python#2

Which I used the "rebase and merge" option, expecting that, since there was no rebase to do, it would just FF merge... not the case, as a new commit was created.
This new commit had coverage uploaded in the push, so also shows up in Codecov for the main branch.

Conclusion:

• I couldn't "properly" make a FF merge (I think) but example 1 does have a merge, and the commit from the feature branch was properly moved to the main branch
• The case that started this investigation was most likely not a FF merge issue, but something else.

Of course I can be wrong, but I'll close this for now.
(ps.: I don't have access to more data to further debug the case that did start this investigation)