Advanced git techniques
16 September 2024
Easy ways to deal with uncommon complex situations in project lifecycle
In contrast with the previous articles, which suggest to fully integrate the daily workflow, this is about the complex situations that may arise only a few times in a project.
There are a few advanced git techniques that can help in more complex situations. The best thing is, you don’t even need to really know how to come up with the solutions, just to know how to look for them.
A cautionary tale
In one company I worked, there was a day when pulling the master branch took a long time. A very long time. For security reasons, we had to work on the code through ssh on a rather slow machine, and so we were advised not to pull while they sorted out the situation.
What happened was that the tests generated hundreds of megabytes of data in the project folder. Somebody committed all that to master, and pushed apparently without noticing anything (was he using an IDE?). By the end of the day, they had sorted out the situation. The project lead created a new repository from a zip archive of the day before, and deleted the old one, effectively wiping out the whole history.
You see, reverting the commit means that the commit is still in history, so the repository will stay with hundreds of megabytes forever. Right?
There are so many wrongs about this that it should have been painful to even read.
First, the solution was to force-push the branch, notify everyone to delete their local branch and check it out again.
That would take maybe 5 minutes of each collaborator’s time, considering the time to cherry-pick their commits if they work on master with a “rolling rebase” approach, the bulk of time being spent on reading the email.
Git hooks could have been put in the project to avoid the situation repeating.
Going back to the source, the test data folder that was created should have been put in the .gitignore.
To this day, I cannot fathom how a team could do something like this. The project lead was in another country and notoriously in bad terms with the lead of my own team, so we were told we had to roll with it.
Get all files changes affected by a feature branch
This one can be quite useful to optimize the test suite in case it is very long. Instead of running all the tests, you can run only the ones that are affected by the changes. The workflow becomes:
- run the affected tests on push
- run the full test suite on merge
- run the full test suite manually in doubt
With this command, you can get all the files that were changed by a feature branch:
git diff --name-only master...feature
Actually, the hard work is in configuring the CI pipeline properly.
Import history from another repository
When merging projects or migrating, you may want to include history from another repository.
This (regularly edited) discussion provides different methods for this: Stack Overflow - How to import existing Git repository into another?
Cherry-pick all commits affecting a given folder
The following command displays the commit history for a specific folder (or path) in a repository:
git log --oneline --name-only --follow -- path/to/folder
From there, cherry-pick those commits, for example in a bash script:
for commit in $(git log --oneline --name-only --follow -- path/to/folder); do
git cherry-pick $commit
done
Overwrite the author in all the git history
Sometimes, you want to change the author of all the commits in a repository. There can be a variety of reasons for this, but the most common one is that the email address was wrong. This can be done with the following command:
git rebase -r --root --exec "git commit --amend --no-edit --author 'New Author Name <email@address.example>'"
Stack Overflow - How do I change the author and committer name/email for multiple commits?
Complicated situations have a simple solution
With git, there is never a (good) reason to panic. Git internals are complex and the commands are not always intuitive. But whatever the situation, there is always a (reasonably simple) way to solve it.
-- len