I've been tasked with improving our development pipeline and processes since it looks like there are tons of unmerged branches in our repositories. Surprisingly, a lot of this code is still in use! I could use some advice on how to effectively reconcile these branches and set up boundaries to prevent similar issues in the future. Has anyone dealt with a situation like this before? Any insights would be appreciated!
4 Answers
Sounds like you've got a cleanup mission ahead! One approach we used was to write a script that goes through branches and deletes the ones that are stale or fully merged. Just be cautious about that 'code in use but not merged' situation—you can't just ignore it!
Exactly! It definitely complicates things when there's no proper merging.
You might want to set up part of your CI/CD flow to enforce PRs back to the main branch once development is done. It's also useful to run a report across your repos so you can identify which ones have a lot of branches. Cleaning up those older, unused branches will help a ton.
Yes! Getting a report sounds like a solid plan. It helps to see the bigger picture.
For sure. Setting those permissions can make the process smoother as well.
It's wild how there could be branches in use without being merged. I'd suggest auditing the branches with the development team to get a clearer picture. Start by talking to the engineering lead about the mess. It might be a good move to implement something like Gitflow or GitHub Flow for better management going forward.
For real! It’s crucial to establish proper PR practices.
Totally agree! Setting up a process helps everyone stay on track.
In cases like this, I think the key is figuring out how to merge and clean up without losing any important code. If branches have been around for too long, it might help to diff between your production builds and the trunk to identify what's missing. It's not perfect, but it can guide you in the right direction.
Great point! Even if it's not the exact original code, you're right—it's usually close enough to work with.
Definitely. Reviewing that diff could save a lot of headaches later.
Right? The scary part is figuring out how the code got to production without being on the main branch.