I have a server on which I have a bare repository for pushing. However, my server needs to have a working copy of the master branch.
How do I get a working copy and that only from a bare repository?
I have a server on which I have a bare repository for pushing. However, my server needs to have a working copy of the master branch.
How do I get a working copy and that only from a bare repository?
You can use 'git show' for this.
http://www.kernel.org/pub/software/scm/git/docs/git-show.html
Basically:
This is how it works:
Voilà!
For info:
git init
will report:Reinitialized existing Git repository in /path/to/existing-bare-repository
. But be confident.man git-init
says: Running git init in an existing repository is safe. It will not overwrite things that are already there.The magic is that
git init
alone does not make your files appear in the working directory. You have to checkout the root directory.This is a riff off the other 2 answers but it fills the gap for my use case -- it works for updating the repo from the origin and checking out branches and any git operation because you end up with a normal complete git repo.
After executing these 3 lines of code you can then
git pull
andgit branch
andgit checkout some_branch
and so on because you now have a normal complete git repo connected to your remote repo.A bare repository is just the .git directory of a working directory, and an entry in the local config file. What I did to convert a bare repository into a full one is:
.git
and move all files from the bare repository in there.git/config
file to changebare = true
tobare = false
You could set the Hidden attribute on the
.git
directory on Windows, but not on the files inside the directory.You can simply clone the repository to another directory on the same machine:
The current directory will become a non-bare clone of your repo, and you'll get a checkout of the
master
branch automatically. Then use the usual commands likegit pull
to update it as needed.As a side benefit, this operation is very efficient — if you specify a local directory to
git clone
, git will try to share objects between those two repos using hard links.I was looking for the "detached working tree" approach (as seen here):