Since I am in the process of making most of my programs available on github, I have been re-shuffling the some git repositories. Especially my collection of scripts (small programs for different purposes). Because some of those scripts are private, I don’t want to publish that repository as a whole.

The easiest way to do that would be to just copy the files into a new repository and commit them. But that would throw away the whole commit history of the files in question, which I don’t want to do.

Initially I thought that reposurgeon might do the trick. After practicing with it, I came to the conclusion that it didn’t work all that well for this purpose. To end up with only the files I wanted, I had to expunge all other files. That is a lot of work. So I was using the wrong tool for the task at hand. The fact that reposurgeon crashed a lot when doing this didn’t help.

Since reposurgeon uses the command streams generated by git fast-export, I decided to stick with git’s fast-export and fast-import.

As an example, the way I isolated all scripts related to git into a new repository is shown below.

The following command was used to export the histories of the files in question;

git fast-export HEAD -- *git* kw* update-* > ../gitscripts.fi

Then I created and initialized a new repository and imported the history of those files;

mkdir git-scripts
cd git-scripts/
git init
git fast-import <../gitscripts.fi
git checkout master
echo '*.pyc' >.gitignore
echo '*.pyo' >> .gitignore
echo '__pycache__/' >> .gitignore
echo 'backup-*' >> .gitignore
git add .gitignore
git commit -m "Added gitignore file."