4 ms·
And so is "git push --force". Wait, what? I thought you were supposed to avoid "push --force" when pushing to a public repo?
by tlrobinson 17y ago
And so is "git push --force".
Wait, what? I thought you were supposed to avoid "push --force" when pushing to a public repo?
- jrockway 17y agoIt doesn't break git, but it may confuse your contributors.
- pyre 17y agoI'm thinking that he means, if everyone is doing git pull --rebase then a git push --force won't screwup the people you are sharing with. My manpages still seem to warn against using both commands, but I'm locally running on a version dated 2/09/2009 (I have a cronjob at work that pulls down the latest version of git and compiles it daily). [edit] Maybe by 'do the right thing' he means that it won't overwrite their local history (i.e. the pull/merge will fail because the histories are different)
- jrockway 17y agoAll I know is that I "git push --force" all the time, and it has never caused anyone tracking my repositories to complain or stop sending me patches. I assume this means git handles it fine. (I just tested this, and git does handle it fine. It says "forced update", and everything works normally. The reflog stores the old pre-force-push tree still, so you are free to revert and push -f the "old" tree, if you really feel the need to.) Anyway, "git push --force" has the potential to lose data, but less potential than "rm -rf .". You will destroy your central repository, not anyone's actual work. ("Destroy your central repository" sounds bad, but isn't. The important copy is the one on your disk.) A clean history is better than never using "push -f" IMHO. The manual authors apparently disagree.
- pyre 17y ago> The manual authors apparently disagree. Pushing doesn't always mean to a central repository. Since there is no distinction, you could be git pushing while on a branch that is tracking a topic branch in someone else's local (local to them) repo.
- jrockway 17y agoPushing to non-bare repositories is always considered harmful.