GitHub etiquette for a bugix fork that got out of hand? (lemm.ee)

submitted 20 hours ago* (last edited 20 hours ago) by Lojcs@lemm.ee to c/opensource@lemmy.ml

8 comments fedilink hide all child comments

I've forked an app I use myself to add what I thought would be a (relatively) small fix. I then realised it was sort of left in the middle of a redesign/rewrite caused by dependency updates changing / breaking things and decided I needed to finish that if I wanted to see my code on a release. Along the way I got carried away and now the fork feels too large to pull request.

The changes I made don't meaningfully change the app structure but they required multiple refactorings/rewrites and extrapolation of what the redesign was trying to do.

The original developer is active but haven't committed to the repo in a long time. What should I do? I just want people to use the code I wrote. Do I PR and see what they say? Do I make another branch of its main and try to add my changes in more digestible chunks? Do I change the package name and release a build myself? I don't want to figure out a new name / logo etc, especially since the app is largely the same.

There are issues on the original repo about bugs I fixed. I'd like to point them to my fork, but that feels like shilling an unnecessarily made fork. And it doesn't feel right to drop a huge PR on the dev and expect them to accept it or even reply neither.

What do?

top 8 comments

sorted by: hot top controversial new old

[-] sik0fewl@lemmy.ca 8 points 10 hours ago

Create a PR and describe the situation. Ask if they'll accept the PR as-is or if they'd accept the reactors separately.

[-] hperrin@lemmy.ca 1 points 2 hours ago* (last edited 2 hours ago)

^ This

Gives the dev both options and shows you’re not trying to give them a bunch of new work, but do needed work for them.

[-] catloaf@lemm.ee 18 points 17 hours ago

If you can, fork it again but this time only fix the one bug with minimal changes and submit that as a PR. Then ask the maintainer about submitting more to fix the other bugs, and doing some refactoring.

[-] isVeryLoud@lemmy.ca 4 points 4 hours ago

This respects the PR atomicity principle.

This should be done anyway. One commit per fix, one PR per issue.

[-] propter_hog@hexbear.net 4 points 16 hours ago

Yep, this is the way. A PR should have a single focus. So your updates may end up being multiple PRs.

[-] listless@lemmy.cringecollective.io 33 points 20 hours ago

Give it to the dev, and explain the situation. Let them know it seems too big/complex of a PR, but you're willing to make additional changes, or break up PR to make it more palatable for merging. It's in the dev's hands after that.

I don't think you should release your own fork without at least trying to work with the original project.

Make sure all your commits have detailed commit messages so that the dev can follow what you were doing (upgrading deps, refactor because xyz, etc.) Don't just record what was changed, explain why it was changed.

anyway that's what I would do.

[-] doeknius_gloek@discuss.tchncs.de 15 points 20 hours ago

I'd say ask the original developer directly. Getting your changes merged upstream should be the preferred option for you, the original dev and the users. If everything goes right, you both could figure out a way to do this, maybe by re-introducing your refactorings and fixes one by one in smaller pull requests. Maybe you could become a maintainer in the process and support the original dev long term so everybody wins.

If the original developer doesn't respond or declines you could think about bringing your own fork forward. Think about the consequences though, the original dev might get frustrated by a competing fork and abandon the project completely. The users on the other hand might be confused or insecure about which version to choose. Your fork must offer a lot for them to jump ship and switch.

Generally I'd say open source is about working together, not against one another, so just shoot them a message and see where it goes.