Git

Topic overview


Relevant course modules


Key commands

Command Explanation Examples
add

Stage files to be committed, including previously untracked files.

Therefore, git add does two things at once for untracked files: start tracking them and stage them. Files that were already being tracked are merely staged by git add.

 Stage a single file:
git add file.txt
Stages all eligible files, tracked and untracked:
git add --all
Stage shell scripts anywhere in project:
git add *.sh
commit Commit all currently staged changes to create a snapshot for the repository to which you can always return later on. git commit -m "Started the book"
git commit -am "Fix infinite loop bug"
    -m specify commit message as above
    -a Also stage all changes (but won’t add untracked files)
status Get the status of your repository: which files have changed since the last commit, which untracked files are present, etc.
log See the repository’s commit history. git log
One-line summary for each commit:
git log --oneline
Show all branches in “graph” form (useful with >1 branch):
git log --oneline --graph --all
diff

By default, show changes between the working dir and:

  • The stage (index) if something has been staged.
  • The last commit if nothing has been staged.
 Last commit vs second-to-last commit - full repo:
git diff HEAD HEAD^
Last commit vs a specified commit - specific file:
git diff HEAD d715c54 todo.txt
Show changed between stage and last commit:
git diff --staged
mv Move files that are tracked by Git: better to use git mv than regular mv so Git will immediately register what happened properly. git mv old.txt new.txt
rm Delete files that are tracked by Git: better to use git rm than regular rm so Git will properly register what happened at once. git rm tmp.txt
branch Various functionality for Git branches, such as creating, removing, and listing branches. (To switch between branches, use git checkout.) List existing branches:
git branch
Create a new branch named “my-new-branch”:
git branch my-new-branch
Rename the current branch to “main”:
git branch -M main
Remove a branch that is no longer needed:
git branch -d fastercode
checkout

git checkout has multiple functions:

  • git checkout <branch-name> will switch to the specified branch.
  • git checkout -- <file> will revert the specified file back to its last committed state.
  • git checkout <commit-id> will move you to the specified commit for looking around.
 Move to the master branch:
git checkout master
merge Merge a specified branch into the current branch. When doing a merge, make sure that you are on the branch that you want to keep, and from there, merge the branch whose changes you want to incorporate. Merge branch fastercode into the current branch:
git merge fastercode -m "my msg"
remote Interact with “remote” counterparts of repositories: in our case, GitHub repositories. List the repo’s remote connections:
git remote -v
Add a remote (syntax: git remote add <nickname> <URL>):
git remote add origin git@github.com:me/my.git
push Upload (changes to) your local repository to an online counterpart, in our case a GitHub repository. Push currently active branch to default remote connection:
git push
When pushing a branch for the first time, use the -u option to set up an “upstream” counterpart:
git push -u origin main
pull

Download (changes from) an online counterpart to your local repository (in our case a GitHub repository) and merge (as in git merge) the changes into your local repo.

May result in a merge conflict if you are collaborating with others on the remote repo.

 Pull from currently active branch in remote repo:
git pull
clone Download an online repository. After this git clone, you will have a dir CSB in your current working dir with the full repo (incl. history):
git clone https://github.com/CSB-book/CSB.git
tag Add a “tag” to a commit, for instance to mark the version of the scripts used in the analysis for your paper.
ddddddddddddddddddddddddddddddddddddd		 git tag -a v1.2.0 -m "Publish version"
Tags need to be explicitly pushed like so:
git push --follow-tags


Miscellaneous

Code / concept Explanation Examples
module load git Load the OSC Git module. If you forget to do this, Git will still work but you’ll be using a very old version.
.gitignore

A file that tells Git what (kinds of) files to ignore, i.e. files that are not to be listed among “Untracked files” and will not be added with git add --all.

  • .gitignore is usually in the top-level repo dir (recommended!).
  • Don’t forget to add and commit the .gitignore file.
  • Note that files won’t be “retroactively ignored”: files that are already in the repo would need to be explicitly removed.
 Example entries:

Ignore everything in the dir “data”:
data/
Ignore files ending in ~ in the entire repo:
*~
Ignore gzipped FASTQ files in the entire repo:
*.fastq.gz
HEAD A pointer to the most recent (last) commit on the current branch. Note that as a pointer, HEAD can move and will do so e.g. if you switch between branches.
HEAD^, HEAD^^ A Pointer to the second-to-last and third-to-last (and so on) commit. See figure below for details.
HEAD~1, HEAD~2 A pointer to the second-to-last and third-to-last (and so on) commit: easier than ^ as you get further back, e.g. HEAD~8. See figure below for details.
ddddddddddddddddddddddddddddddddddddd
ddddddddddddddddddddddddddddddddddddd


Tips and tricks

Explanatory figures


Advanced (bonus)

Undoing changes and viewing the past

I need to Use this Examples
Undo unstaged changes to a file

git checkout -- <file>

Note: for the most recent Git versions (not yet on OSC), the recommended method is the new command git restore.

 After accidentally deleting README.md or overwriting it instead of appending to it:
git checkout -- README.md
Unstage a file

git reset HEAD <filename>

Note: for the most recent Git versions (not yet on OSC), the recommended method is the new command git restore --staged.

 After accidentally staging README.md which was not supposed to be part of the next commit:
git reset HEAD README.md
Undo staged changes to a file

git checkout HEAD -- <filename>

Note: this irrevocably discards the non-committed changes (your data is only safe with Git once it has been committed!).

 After accidentally deleting README.md or overwriting it instead of appending to it, AND staging these changes:
git checkout HEAD -- README.md
Undo ALL unstaged changes (all files) git checkout -- . After making edits to several files that you all want to get rid of, to revert back to the last commit:
git checkout -- .
Undo ALL staged and unstaged changes (all files)

git reset --hard HEAD

This will take you back to the state of your repository right after the last commit.
View the state of my repo for a past commit

git checkout <commit-id>

This is only for “looking around”. If you decide you want to go back to your repo as it was in an earlier commit, use git revert or git reset (see below).

 git checkout 4dce25f
git checkout HEAD^^
To go back, assuming you’re on branch master:
git checkout master
Undo one or more commits

To undo one or more commits, i.e. to roll the state of your repository back to how it was before the commit you want to undo, there are two main commands:

  • git revert: Undo the changes made by commits by reverting them in a new commit. (Safe – does not change history.)
  • git reset: Delete commits as if they were never made. (Unsafe – changes history.)

(git reset HEAD^ undoes changes made by the last commit as it resets to the second-to-last commit.)

 Undo changes by most recent commit:
git revert HEAD
Undo changed by second-to-last commit:
git revert HEAD^
Undo changes by any arbitrary commit: git revert e1c5739

Undo last commit and completely discard all changes made by it:
git reset --hard HEAD^
Undo last commit and put all changes made by that commit in the working dir:
git reset HEAD^
Undo last commit and stage all changes made by it (“peel off commit”):
git reset --soft HEAD^
Retrieve a file version from a past commit
dddddddddddddddddddd		 git checkout --
or
git show
ddddddddddddddddddddddddddddddddddddd		 Retrieve from second-to-last commit:
git checkout HEAD^^ -- README.md
Retrieve from arbitrary commit:
git checkout e1c5739 -- README.md
To not change after all and go back to the current version:
git checkout HEAD -- README.md

Look at the version from the last commit: git show HEAD:README.md
Revert to arbitrary earlier version using redirection:
git show ad4ca74:README.md > README.md
ddddddddddddddddddddddddddddddddddddd


Merge conflicts

A merge conflict can occur when all three of the following conditions are met:

When this occurs, Git has no way of knowing which changes to keep, and will report a merge conflict. When it does so, your merge process has basically been paused, and Git wants you to make changes to resolve the conflict, after which a git commit will complete the merge. To resolve a merge conflict:

  1. Use git status to find the conflicting files.
  2. Open and edit those files manually to a version that fixes the conflict. Note that Git has added <<<<<<<, =======, and >>>>>>> markers to help you see the conflict – but you will want to remove those.
  3. Use git add to tell Git you have resolved the conflict in a particular file.
  4. Use git status to check that all changes are staged, at which point Git should tell you “All conflicts fixed but you are still merging. (use "git commit" to conclude merge)”.
  5. Use git commit -m "My merge message" to conclude the merge.


Forking and Pull Requests

Forking is a GitHub concept that will create a personal GitHub repo that remains linked to the source repository. For instance, you could keep your fork up-to-date with the source, and you could issue a Pull Request from your fork. See these slides for details and screenshots.


Miscellaneous

Code / concept Explanation Examples
stash

git stash temporarily saves files.

It can be useful when you need to pull from remote, but are prevented from doing so because you have changes in your working dir (and those are not appropriate for a separate commit / new branch). In such a case, stash the changes, pull, and get your stashed changes back (see example on right).

 Stash changes to tracked files:
git stash (Add -u to include untracked files)
Pull from the remote repository:
git pull
Apply stashed changes:
git stash apply
commit --amend

Add something to a commit that has already been made, for instance if you notice a type or forgot a file.

Note: amending commits “changes history” and as such is not recommended by everyone – but at any rate, do not amend commits that have been pushed online!

 git commit --amend --no-edit (No edit will keep the same commit message)
checkout -b Create a new branch and move to it immediately. git checkout -b my-new-branch
git log <filename> Show commits in which a specific file was changed (!).
ddddddddddddddddddddddddddddddddddddd		 Say you have a script that no longer works and you need to see the changes that were made to it:
git log my-failing-script.sh


Reuse

Text and figures are licensed under Creative Commons Attribution CC BY 4.0. The figures that have been reused from other sources don't fall under this license and can be recognized by a note in their caption: "Figure from ...".

Jelmer Poelstra, 2020-2021