Clean your Git history
When I started my .dotfiles repo it wasn’t very obvious to me how to keep my atom packages in it. As a consequence I just decided to push into my .dotfiles everything contained in the
.atom directory under my user directory on my machine.
It was an easy and no brain solution that worked for a while but my repository and commits became hard to work with because of the numbers of files contained in every commit (+15000 for some commits). So I had to look for a solution to clean the repository retroactively.
Remove private things (password/keys) from a Git repository
This technique can also be used to remove something that you did push in a repository which shouldn’t be there. Passwords and API keys are good clients for those mistakes. Sometimes they are pushed inadvertently and of course you should remove them from your repository right away.
As explained here by SXX, a good practice is also changing these passwords and API keys. Even they won’t be visible in the future that doesn’t mean that they haven’t been seen while you pushed them.
Only remove from the history
This version supposes that you already have
.gitignore the things that you want to remove from your repository. If you haven’t, just go to the next section.
git filter-branch --tree-filter 'rm -rf atom.symlink/packages' --prune-empty HEAD git gc git push origin master --force
Remove from the history and gitignore
Let’s suppose you have just pushed a
config.js containing passwords and you want to remove it.
git filter-branch --tree-filter 'rm -rf config.js' --prune-empty HEAD echo config.js >> .gitignore git add .gitignore git commit -m 'Remove config.js from Git history' git gc git push origin master --force
For the curious, what is
If you are like me you are now wondering what is the
git gc command. Is there a garbage collector in Git? Almost indeed ;)
git-gc - Cleanup unnecessary files and optimize the local repository. From Git documention
If you look at the description on the same site you will read that: “Runs a number of housekeeping tasks within the current repository, such as compressing file revisions (to reduce disk space and increase performance) and removing unreachable objects which may have been created from prior invocations of
I strongly encourage you to go to the git documention and read the rest of the description.