After merging 2 branches, Git randomly decided not to merge one particular line (the place where my newly defined function was called) and that caused a fixed bug to reappear. First time in 4 years I am witnessing Git do something strange like this— probably an issue in the “merge by ort strategy”.

git gud

After a merge I like to go through every file where conflicts were fixed and read the conflicting functions top to bottom.

I've seen that too. Buggy code suddenly makes a comeback. Fixes put in the code disappear out of nowhere.

Most think a merge works like this:

* commit 1 ... commit N from branch A
* iterate over each commit
* try to apply in branch B

It's not entirely wrong, but there is an important detail:

merge needs to find a common ancestor for the two branches A and B.

https://git-scm.com/docs/git-merge/...

As written here...

This is what git merge-base does:

https://git-scm.com/docs/...

Especially in merging large histories, it's a good idea to validate that the ancestor matches the expected one.

Another good tip: Check if the branches history was rewritten or contains reverts .

Having a close look at what commits are merged can prevent some funky results.

E.g. merging reverting commits should be double checked, cause it can happen that the revert commit undoes correct changes :) sounds like what some here described.

This is why I like rebasing before merging.