[Discussion] Do we want to keep our fork in sync with upstream?
Upstream has a bunch of commits we don't have, and there was one commit we had that wasn't upstream (94131e16; I just sent it upstream with slight tweaks to the title as https://gitlab.com/marge-org/marge-bot/-/merge_requests/413).
Do we want to:
- keep this revision and never update
- only backport select commits, diverging from upstream
- sync every commit, but in batches that get reviewed before merging
- automatically pull from upstream so that we have everything
Note that we auto-deploy on push to main
, so option 4 means if something is merged upstream and breaks our use case, all we can do is send a fix upstream and wait for it to be merged.
I'm in favour of option 3, where we keep our git history in sync (same commits, not just the same changes in new commits), but we check everything first so that if something can cause issues we have a chance to catch it, and if a break does happen we can revert main
back to an earlier commit and only sync the next batch once upstream includes a fix.