Uh oh!
There was an error while loading. Please reload this page.
stop recommend rebase for new contributors - #496
Conversation
We should not recommend "rebase" and "push -f" to new users.
asvetlov
left a comment
There was a problem hiding this comment.
I think it makes sense.
We squash on merging anyway
| you run ``git merge upstream/master``. | ||
| When it happens, you need to resolve conflict. See following articles for | ||
| how to resolve conflicts. |
There was a problem hiding this comment.
Could anyone check my English?
- happens vs happened
- "happens, " -- this comma is usual?
- "for how to..." -- is this usual wording?
There was a problem hiding this comment.
You're using "happens" correctly.
The common after "happens" is a typical usage.
I'd say something more like "See these articles about resolving conflicts:". The "for how to" is a little awkward.
Mariatta
commented
Jun 13, 2019
methane
commented
Jun 14, 2019
Would you please link? I want to confirm what happened there. |
willingc
commented
Jun 15, 2019
It may be better to recommend to new contributors that they should be developing on a branch in their account for each feature. As for rebase, I rarely use I don't object to adding a note for new contributors about rebase though. |
methane
commented
Jun 16, 2019
git rebase doesn't reduce conflict. It even increase conflict sometime. |
willingc
commented
Jun 16, 2019
@methane The explicit git fetch/git rebase combined with specific remote naming, branch usage, and update ordering approach that I teach does reduce merge conflicts significantly. However, I do agree with you that using rebase when not following the particular workflow that I use can have unexpected results. I recommend a few things for this section:
As @asvetlov mentions, it doesn't really matter since we squash. |
terryjreedy
left a comment
There was a problem hiding this comment.
If I have ever used rebase, it was because my fork and local repository was ahead or beside upstream. For normal updates when behind, I fetch and merge to each workspace, and to PR branches when working on them. I only get merge conflicts in idlelib when an old PR branch touches code affected by a different merged PR.
In other words, the revised doc matches what I do and have done successfully, as a git non-expert, for 2 years.
willingc
commented
Jun 16, 2019
Good to hear @terryjreedy. I'll approve this PR as stands and add the other notes in a separate PR. |
FWIW, I leave two comments about why I think merge is better. If the guide is for contributing to general OSS, it is recommended to keep commits in the pull request branch clean. That's why many people recommend rebase. But we're using "Squash and merge" workflow, cleaning commits in pull request branch is not so important. Keeping history clean is complex than just using merge. When people are expert and keep history in the branch clean with There are many documents about |
methane
commented
Jun 17, 2019
And github creates much noise when |
methane
commented
Jun 18, 2019
@Mariatta I found the issue you reported. And it may be caused by rebase workflow! After 3.8a1 is released, ma8ma creates different commit object (see "ma8ma committed" in the history). Maybe, Anyway, merge workflow doesn't create such "rewrote" commits. |
Mariatta
commented
Jun 18, 2019
Huh, very interesting.. Alright. Thanks for following through with this, @methane. |
* stop recommend rebase for new contributors We should not recommend "rebase" and "push -f" to new users. * fix English and rest syntax



We should not recommend "rebase" and "push -f" to new users.
These command requires high skill when they cause trouble.