git ready

learn git one commit at a time
by Nick Quaranto

push to only bare repositories

committed 01 Feb 2009

Typically Git allows you to build whatever kind of workflow you’d like. Instead of saying “you must use it this way”, Git lets you figure out what best works for you and your organization. Like every system though, there are caveats and gotchas, and this tip goes over one of them. Let’s go over some definitions about the subject first:

bare repository: a repository cloned using the --bare option, only includes the files/folders inside of the .git directory
non-bare repository: a normal clone, has a working directory with checked out files

There’s easy ways to share your changes with Git, but sometimes you want to push and pull from a repository instead. The best practice here is when you want to push changes, don’t push to a non-bare repository. In other words, from the GitFaq:

A quick rule of thumb is to never push into a repository that has a work tree attached to it, until you know what you are doing.

The lesson here is that if you’re going to push changes, make a bare repository clone with git clone --bare. From this point, you could host the changes from with git daemon so others can access them. (Tips on this command will be coming!) The reason why Git was built to be unhappy with pushing to non-bare repositories is because one of its main goals is to not lose your data.

If you do want to for some reason, measures are in place so you don’t accidentally lose your changes. Once you have pushed to a non-bare repository, you’d have to git reset --hard HEAD to throw your changes out, and then do a git checkout -f to bring the pushed changes in. Even if you did lose some work that was committed, you could use git reflog to pull the objects back out of Git’s datastore.

Now that you know what it takes to push to a non-bare repository, hopefully you’ll take the easier route and just use the --bare option. If for some reason you don’t want to, now you know what can be done to work around it.