This might raise some questions for fervent Git fans: Why would anyone want to do this? Why not just keep everything in Git?

Here’s a personal anecdote: I developed a plugin for WordPress called GitPress. My initial development was done on GitHub, but in order for my plugin to be available via the WordPress Plugins Directory, I needed to bring everything from GitHub into WordPress’ SVN repository.  Here’s how I did it:

First, we create the initial SVN remote.  By using the standard layout (--stdlayout) and specifying the “gitpress/” prefix, I can easily get tracking branches for SVN branches, tags, and trunk.

git svn init --stdlayout --prefix=gitpress/ --username=enricob http://svn.wp-plugins.org/gitpress

Now, I fetch everything from the SVN remote that I’ve created and see if the remote branches are there:

git svn fetch

Now, when I list my remote branches in Git, I see gitpress/trunk, gitpress/tags, and gitpress/branches.  I create a local tracking branch for gitpress/trunk, which is where I’ll be pushing my changes:

git checkout -b svn-trunk gitpress/trunk

Now that I’m in my new tracking branch, I bring in the changes from my Git repository’s master branch:

git merge -s subtree master

Finally, I can commit my changes to SVN in one squashed commit:

git svn dcommit

I’ve also found that using git svn rebase causes separate commits to be made instead of getting a squashed commit, but the cases I’ve run into are pretty simple so far.

So that’s how I brought my GitHub project into an SVN repository. Hopefully this will help somebody else who might run into the same situation.