Non-semilinear repository history #182
Replies: 5 comments 4 replies
-
Uh, I think this is actually what the intended Bors result is 😢 (see here, with discussion here). |
Beta Was this translation helpful? Give feedback.
-
I did an extensive test of the auto-merge feature in GitHub, and the issue is that if we want to ensure that the master branch is always tested before merging, we essentially have to tell GitHub to require that branches are up-to-date before merging, and it won't automatically rebase them, so while we as maintainers end up having to merge all branches in sequence, one after the other manually. So that's not optimal, especially if we ever end up with more frequent other contributors. |
Beta Was this translation helpful? Give feedback.
-
Yeah. Shame! We were applying a semi-linear capable workflow at my company; we ultimately didn't - I think it's possible but tedious to setup. I'm not fussy about history outlook - it would have been a nice to have, IMO. Granular commits are definitely very important for me 😁 Last bisection was while developing the stages PR, which actually helped isolating an issue 😍. Yeah, I'm a big fan of granular commits!! 😂 |
Beta Was this translation helpful? Give feedback.
-
I think I may have found a bot that will do what we need: https://github.com/palantir/bulldozer I'm still investigating. |
Beta Was this translation helpful? Give feedback.
-
I'll have to test drive this one: https://github.com/chdsbd/kodiak It touches the PR branch to update it with master, but only once you tell it to merge, so that's probably fine. I want to test it out first to make sure it feels right, though. |
Beta Was this translation helpful? Give feedback.
-
I've noticed that the repo history is not semi-linear:
I thought that Bors would have generated something like:
am I correct (I've never used Bors)?
If so, any idea why?
Beta Was this translation helpful? Give feedback.
All reactions