Skip to content

Latest commit

 

History

History
807 lines (687 loc) · 36.7 KB

git.md

File metadata and controls

807 lines (687 loc) · 36.7 KB

GIT cheatsheet

Overview

Location Transition Command
working directory git add
staging area git commit
local repository git push
remote repository "Pull Request on Github"
upstream repository /

Many git commands (e.g. git diff, git log, git show, etc.) show their output in the less pager.
You can exit less with the q hotkey, and search it using / followed by the search term;
n and N move forwards/backwards in the search results.

Tip:
git [command] --help will give you information on any git command.

Basic git commands and usage, alphabetically:

$ git add [filename]

Add a file to be tracked by git; it is now said to be staged.
(It still needs to be committed to actually be included in version control)

  • git add .
    Add all files in the directory. This is useful if you want to upload an existing project to Github.
    (Use a .gitignore file that lists all exceptions)

$ git blame [path/to/file]

This will show you the commit hash, the time and the committer for each line.
Then you can use the commit hash in git log [hash] or git show [hash]
to find out more about it, like why it's done this way (if documented).
You can also use the -L flag to indicate the from-to line numbers you want info about:
git blame -L 38,41 src/color/adjustHue.js
Or for a single line, use:
git blame -L 38,38 src/color/adjustHue.js
Note:
git blame only reveals the last introduced changes to each line, not who or when the lines
were first introduced into the codebase. This can be rather annoying if the last change
was for example an indentation change.
Use git log -S to find the commits that first introduced a specific line of code.

$ git branch [branch-name]

Create a new branch.

  • git branch -a
    Show all branches.

  • git branch -u origin/[remote-branch]
    = git branch --set-upstream-to=origin/[remote-branch]
    This sets the default remote branch for the current local branch.
    Any future git pull command (with the current local branch checked-out), will attempt to bring in commits from the 'remote-branch' into the current local branch. One way to avoid having to explicitly do --set-upstream-to is to use the shorthand flag -u along with the very first git push as follows:
    git push -u origin local-branch
    which sets the upstream association for any future push/pull attempts automatically.

$ git checkout cffd76070aaebf82e4b5eb330fe9a2df944c1b81

Checkout commit cffd76070aaebf82e4b5eb330fe9a2df944c1b81.

  • git checkout master (--force)
    Move back to the master/head commit.
    --force: When you've made changes that aren't committed, git won't let you checkout another branch.
    The "--force" flag forces git to checkout anyway, which will result in losing the changes.

  • git checkout -b test_branch (cffd76070)
    = Shorthand for:

    git branch test_branch (cffd76070)  
    git checkout test_branch
    

    (cffd76070): Optionally specify a commit hash to base the branch on. When not specified, it is based on the currently checked out commit (current position of head) So when you're in detached head state (= after checkout of commit hash instead of branch), you can simply run 'git checkout -b test_branch' to turn that commit into a real branch. Afterwards, simply:

    git checkout master
    git branch -d test-branch
    

    to checkout master, and delete the branch (locally) if you don't need it anymore.
    To also delete the branch remotely (on Github):
    git push origin --delete test_branch

  • git checkout --orphan [branch-name]
    Create a completely empty branch, e.g.:

    git checkout --orphan gh-pages
    git reset --hard
    git commit --allow-empty -m "Initialize gh-pages branch"
    git push origin gh-pages
    git checkout master
    

$ git cherry-pick (-x) <HASH>

Cherry picking in Git means to choose a commit from one branch and apply it onto another.
This is in contrast with other ways such as merge and rebase, which normally apply many commits onto another branch.
Cherry picking can also be used to restore deleted commits from the reflog.
-x: This generates a standard commit message with mention where it was cherry-picked from.
This is good practice when cherry-picking from a public branch.

  • git cherry-pick -x 0379e861ca82^..bb2197f0845d
    Cherry-pick a range of commits to include in the current branch.
    Note:
    • Both hashes should be from the same branch, with the oldest specified first.
    • Include ^ at the end of the first hash if that commit itself should be included;
      without it, the range will start at the commit following the specified one:
      0379e861ca82^..bb2197f0845d = [0379e861ca82, bb2197f0845d]  
      0379e861ca82..bb2197f0845d  = ]0379e861ca82, bb2197f0845d]
      

$ git commit (-a) -m "Description of changes"

Commit all staged files permanently to version history.

  • git commit -am "Description of changes"
    -a/--all: Automatically stage/add all tracked files that were modified; this means you don't have to manually stage/'git add' all files that are modified.

  • git commit --amend
    Rewrite the most recent commit message and/or commit content (if they have not yet been pushed online).
    If you want to also edit the commit's content, you need to 'git add' the edited files before amending,
    so you may or may not have to 'git stash' code that shouldn't be included.
    (In Git, the text of the commit message is part of the commit.
    Changing the commit message will change the commit ID, i.e. the SHA1 checksum that names the commit.
    Effectively, you are creating a new commit that replaces the old one, but keeps its date.)

$ git diff

Shows all modifications that are NOT yet staged.
(= difference between working directory and staging area)
Note:
git diff output is shown in the less pager (q to exit).
To search in the less pager, use / followed by the search term and press enter.
Use n and N to move forwards/backwards in the search results.

  • git diff --staged
    Shows all modifications that are staged.
    (= difference between staging area and last commit)

  • git diff cffd76070aaebf82e4b5eb330fe9 1e14e1134d7eabf136d6210a6d87a99
    or: git diff cffd76070aaebf82e4b5eb330fe9 1e14e
    or: git diff cffd 1e14
    = git diff [old] [new]!
    (you may abbreviate commit hashes; the first 4 or more chars suffice)
    Compare 2 specific commits (get commit hash from 'git log' command)

  • git diff HEAD~ HEAD
    Show the difference between the head commit (= HEAD = HEAD~0) and the one before (= HEAD~ = HEAD~1).
    In other words, it is identical to the simpler git show command.

  • To permanently exclude a specific file in your repository from the diff command:
    git config diff.nodiff.command /bin/true
    cat >.git/info/attributes
    > package-lock.json diff=nodiff
    From now onwards, package-lock.json will be excluded from any diff.
    Use --global in the first command to accomplish this for all your repos.

$ git fetch origin

Get the entire repo from the origin remote, including all branches, without actually committing them to your local repository. Instead, git fetch stores the fetched files in .git/refs/remotes. You can now either merge or rebase your local branch.
In git, a ref or reference is simply a file stored in .git/refs, containing the hash of a commit object, e.g. a branch or tag. They allow us to refer to branches with convenient names instead of commit hashes.

$ git log

Returns a log of all git commits in the less pager (q to exit).
To search in the less pager, use / followed by the search term and press enter.
Use n and N to move forwards/backwards in the search results.
Alternatively, you can also specify a commit hash:
git log 0a6bb439280d77458f
This will show the specified commit hash at the top with its parent commits below it. In other words, its child commits (commits made afterwards) are ommitted.
To print the git log output to the terminal instead of viewing it in less, use:
git log | cat

  • git log -L 3,4:[path/to/file]
    This command will present you with a log of all commits that changed line 4 in the specified file.

  • git log -p -- [path/to/file]
    This command allows you to efficiently view the change history of a single file.
    It displays a log of all commits that changed the specific file, along with their diffs.

  • git log -S 'search-string'
    This gives you the commits in which the 'search-string' was introduced.
    Be aware that this command will return every commit in which this 'search-string'
    was added to any file.
    To find when a specific line of code was introduced in a specific file, use:
    git log -S ['Full-Line-of-Code'] [path/to/file]
    This gives you the commit(s) in which this line of code was introduced.

  • git log --format=fuller
    Shows the difference between AuthorDate and CommitDate. They are usually the same, but can differ for example after a 'git rebase'. The CommitDate is taken into account when creating the commit hash, so a 'git rebase' without changes still modifies the hash.
    Other 'git log'-formats: oneline, short, medium, full, fuller, email, raw, ...

  • git log --name-status
    Adds the names and status of the changed files of each commit to the log.

  • git log -5 --name-status --oneline | cat
    Produces a short printed log of the last 5 commits.

  • git log --stat --oneline --graph master other_branch
    --stat: shows additional data like the files that were changed, and insertions/deletions.
    --oneline: show only one line per commit (see other --format types above at git log --format).
    --graph master other_branch: show a graph of how the commits in both branches relate to each other.

  • git log --reverse
    The first entry in the output will be the first commit, instead of the last one as usual.

$ git ls-files

Lists all the files that exist in the latest commit on the current branch.

$ git merge other_branch

Merge other_branch into the currently checked out branch.
To merge other_branch into master, simply check it out first:

git checkout master
git merge other_branch

The key is to remember that git merge always merges all the specified branches
into the currently checked out branch, creating a merge commit for that branch.
This means the current branch is always included in a merge!

  • git merge --abort
    Restore your files to their state before you started the merge.

$ git mv old_filename new_filename

Rename a tracked file.

$ git push [REMOTENAME] [BRANCHNAME]

Pushes the specified branch to the specified remote.

  • git push --set-upstream origin --all
    = git push -u origin --all
    Pushes all branches to the origin remote, and sets origin as 'upstream' for all of them.
    This means on next push you don't need to specify neither your branch, nor
    your remote; simply run 'git push'.

$ git pull origin master

= Shorthand for:

git fetch origin
git checkout master
git merge origin/master

When you clone a git repository from github, you get separate pointers to master (= latest commit in local master branch) and to origin/master (= latest commit in githubs master branch). When you commit locally, only the master branch is updated. When you commit on github and then fetch, the origin/master branch will be updated locally. If you then merge both branches, you'll get the same result as a 'git pull origin master'.
(But if 'git pull' is the same as 'fetch + merge', why don't we get merge commits when pulling?
=> This is because git by default performs 'fast-forward merges' when one commit is reachable from the other; the oldest commit's label is fast-forwarded to the head of the branch. No merge commit is necessary since they are already linked.)
(-> Udacity Git/Github Lesson 4, 21.)

  • Pull request (Github-specific):
    You request the original author to pull your changes into his master.
    -> could also be called "Merge request"
    To contribute to a public project on Github, you have to first 'fork' it (similar to clone, but Github-specific), so you have your own version of the project to make changes to. Then you can start a pull request to have the author of the original project (commonly referred to as 'upstream') include your changes into his repository.
    When contributing to a public repository on Github, it’s standard practice to make the changes in a non-master branch within the fork. This way, you can easily keep your master branch up-to-date with the master of the original repository, and merge changes from your master into your branch when you are ready.

  • Keeping a Github fork up to date:

    • Clone your fork:
      git clone https://github.com/YOUR-USERNAME/YOUR-FORKED-REPO.git
    • Add remote from original repository to your forked repository:
      cd into/cloned/fork-repo  
      git remote add upstream https://github.com/ORIGINAL-DEV-USERNAME/REPO-YOU-FORKED-FROM.git  
      git fetch upstream
      
    • Update your fork from original repo to keep up with their changes:
      git pull upstream master
    • Push the changes to your fork:
      git push

$ git rebase other_branch

Rebase the current branch on the specified other_branch.
This means the current branch's new commits will be reapplied on top of other_branch's new commits;
only the current branch changes, while other_branch remains unchanged.
Similar to git merge, git rebase integrates changes from one branch into another.
The big difference is that merging preserves history whereas rebasing (potentially) rewrites it:

  • Merge feature_branch into master:

    git checkout master
    git merge feature_branch
    • If there are no new commits in master, the merge is fast-forwarded, meaning no new commit is generated.
    • If new commits are present in master, a merge commit will be generated, but no commit hashes will be changed.
    • Master changes, feature_branch doesn't.

    => This is what you should do when the feature is finished, and its ready to reintegrate with the master branch.

  • Rebase feature_branch on master:

    git checkout feature_branch
    git rebase master
    • If there are no new commits in master, the rebase will have no effect whatsoever.
    • If new commits are present in master, the commits from feature_branch will be reapplied on top of them, rewriting the feature_branch commit hashes.
    • feature_branch changes, master doesn't.

    => This is what you should do if the feature is not finished, but relevant changes have happened to master:
          by rebasing you can simply continue working on the feature on top of the new master changes.

The easiest way to undo a rebase is to find the commit hash of the branch as it was before the rebase started
and reset to it. Use git log or git reflog to find it.

  • git rebase -i <HASH>
    -i, --interactive: Interactive rebasing can rewrite git history; unlike git commit --amend it also allows modifying commit messages further back than the last one.
    Examples:
    git rebase -i 'Commit-Hash-You-Want-To-Change'^
    Note the ^ is necessary to refer to the parent of that commit; otherwise you'll be off-by-one!
    git rebase -i HEAD~3
    Change the commit messages or commit content for the last 3 commits.
    This command will present you with your default editor; change 'pick' to 'reword' for the commit message(s) you want to change, then save and close. A new editor window will appear for each commit message you wish to change.
    If you also want to change the commit content, change 'pick' to 'edit', then save and close. Back in your shell, you'll notice you're no longer on 'master', but on the specific commit hash you chose to modify; simply change the code you want, 'git add' the changes, and git commit --amend the commit. Enter 'git rebase --continue' to finish up the rebase. Most likely this will include fixing a merge conflict because git will have trouble auto-merging the succeeding commit with your change ...
    Note:
    Amending the commit message will result in a new commit ID since the message itself is used in the SHA1 hash that generates the commit ID. However, in this case, every commit that follows the amended commit will also get a new ID because each commit also contains the id of its parent.
    So if you have already pushed to GitHub, you will have to force push the amended messages/content.
    However, 'force pushing' is strongly discouraged, since this changes the history of your repository. If you force push, people who have already cloned your repository will have to manually fix their local history. For more information, see "Recovering from upstream rebase" in the Git manual.

$ git reflog

Show the reference logs. The reflogs record when the tips of branches are updated;
this means they contain all commits that are/were created in your local repository.
This is very useful to restore commits you lost after a reset; simply checkout or cherry-pick the removed commit by its hash.

$ git remote add origin [github-repository-URL]

Add a (github) remote to your project. origin is the conventional name for the remote repository that a project was originally cloned from. Afterwards, you can push or pull it by running:
git push origin master
git pull origin master
(Run git remote -v to check if the remote was added correctly)
(To change the URL, for instance from git to https, use git remote set-url)

  • git remote add upstream [REMOTE-REPO-URL]
    upstream is the conventional name for a parent repository, e.g. the parent/upstream of a forked repo.
    When collaborating, it's often necessary to add the upstream repository as a remote, in order to resolve merge conflicts between your fork and its upstream parent. Then you can pull in the conflicting changes from upstream/master:
    git pull upstream master
    and merge the master branch into your 'change' branch:
    git checkout change
    git merge master (-> merges all specified branches into the currently checked out 'change' branch)
    Then resolve the local merge conflicts:

    • Update 'file.txt'
    • git add file.txt
    • git commit (-> the 'merge commit')

    And push your 'change' branch to your fork (NOT upstream):
    (this will automatically update your pull request)
    git push origin change
    Finally, just update the master branch on your remote Github fork:
    git checkout master
    git push origin master

$ git remote set-url origin https://github.com/USERNAME/REPOSITORY.git

Change the origin remote's URL.

  • git remote set-url --add --push origin https://github.com/USERNAME/OTHER-REPO.git
    Add a new push URL to your repository. This is mostly useful for pushing to several remotes at once with a single git push. However, adding a new push URL will replace the default push location, so you'll have to add the original again:
    git remote set-url --add --push origin https://github.com/USERNAME/ORIGINAL-REPO.git
    After this, pushing to origin will push to both remote repos.

    Alternatively, you can keep the original origin remote with a single push URL, and create a new all remote that pushes to all repos:

    git remote add all https://github.com/USERNAME/ORIGINAL-REPO.git
    git remote set-url --add --push all https://github.com/USERNAME/OTHER-REPO.git
    git remote set-url --add --push all https://github.com/USERNAME/ORIGINAL-REPO.git
    

    You can then set it as "default" push with:

    git push -u all --all
    

$ git remote show origin

Display complete info on the origin remote's push and pull configurations per branch.

$ git reset (--hard)

Unstages all staged files.
--hard: Also resets the working directory, so any uncommitted changes will be lost!

  • git reset main.go
    Unstages main.go, but preserves its contents.

  • git reset 6f16fb26b3294 src/app/app.js
    Reset a specific file (app.js) to the state of a previous commit.

  • git reset 6f16fb26b3294 (--hard)
    Undo all commits that came after 6f16fb26b3294!
    --hard: will also get rid of everything you've done since then; all these commits will be lost from the working directory.
    In this case, since they were previously committed, you could still retrieve them using git reflog.

  • git reset HEAD~ (--soft/--mixed/--hard/--keep)
    Undo the last commit from the current branch. To restore it use git cherry-pick with the commit's hash.
    (HEAD is a pointer to the most recent git commit in a specific branch; detached head means the head does not point to the most recent commit. This happens when you move the head back in time by checking out a specific commit hash)

    • --soft: The modifications from the last commit remain in the working directory and are still staged.
    • --mixed: The modifications from the last commit remain in the working directory but are no longer staged.
      This is the default mode.
    • --hard: The modifications from the last commit are deleted from the working directory, as well as any uncommitted changes. Note that it is impossible for git to restore modifications that were never committed in the first place, so be careful using this command!
    • --keep: Basically a safer version of --hard; only updates 'change-positive' files (= files that are different between the specified <commit> (= HEAD~ = HEAD~1) and the current HEAD (= HEAD~0)). It does this by performing a merge. Local changes are kept, but if a 'change-positive' file has local changes, the reset is aborted.

    Note:
    If the commit was already pushed to your remote, you will have to push the newly deleted version again with the --force flag to delete it from the remote:
    git push origin master --force
    Be aware this is a bad practice for public repos, since they might already have been cloned by other people.

$ git revert cffd76070aaebf82e4b

Undo only the changes that happened in this specific commit.

$ git rev-list --count HEAD

Count the commits for the branch you are on.

$ git rm filename

Remove a file from both git and working directory,
so it doesn't longer show up as an untracked file.

$ git show

Show various types of objects, highlighting the most recent modifications.
Without any other options, it defaults to HEAD and shows the last commit:

git show = git show HEAD  = git show HEAD~0 = show the head commit 
                                            = most recent commit of the current branch.  
           git show HEAD~ = git show HEAD~1 = show the commit before the head commit.  

                            git show HEAD~2 = show the commit 2 commits before the head commit 
                                            = third latest commit overall.  
                            Etc.
  • git show cffd76070
    Show commit cffd76070.

  • git show src/app/app.js
    Show only the file 'src/app/app.js' of the last commit (HEAD).

  • git show cffd76070:src/app/app.js
    Show only the file 'src/app/app.js' of commit cffd76070, highlighting the modifications this commit added to this file.

$ git status

Lists all new or modified files to be committed.

$ git stash

Takes the dirty state of your working directory (uncommitted changes in tracked
files) and saves ('stashes') it on a stack of unfinished changes that you can reapply at any time.

  • git stash list
    Lists the entire stash (each git stash command creates a new snapshot).

  • git stash apply
    Apply the most recent stash.

  • git stash apply stash@{2}
    Apply an older stash; get its id from git stash list.
    stash@{0} = the most recent stash
    stash@{1} = the stash before the most recent one
    Etc.

  • git stash drop (stash@{2})
    Remove a stash from your stack.

  • git stash pop (stash@{2})
    Apply the stash and then immediately drop it from your stack.

  • git stash push -m 'Put counter in Status' src/app/status/status.js
    Stash a single file, by specifying a stash message and a path to the file.
    (NOTE: git 2.13 and up ONLY!)

  • git stash show
    Shows which files were changed in the latest stash.

  • git stash show -p
    View the content of the most recent stash (-p: 'patch').

  • git stash show -p stash@{2}
    View the content of a specific stash. If, however, you wish to view the content of a single file from a specific stash, you need to use the git show command:
    git show stash@{0}:src/app/app.js
    If you want to see the diff of a specific file in the stash, you have to use the git diff command:
    git diff stash@{0}^1 stash@{0} -- <filename>
    The stash@{0}^1 shortcut means the first parent of the given stash, which is the commit at which the changes were stashed away. We use this form of "git diff" (with two commits) because we have to tell git which parent we want to diff against. Equivalent, but more cryptic:
    git diff stash@{0}^! -- <filename>

$ git submodule add https://github.com/matcornic/hugo-theme-learn.git themes/hugo-theme-learn

Add the project located at the github link as a submodule to the project in the working directory. This allows combining several git repos in a single project.
Beware:
When pulling a github project, its submodules are NOT automatically pulled along!
For that, you have to:
git submodule init
git submodule update
If you want to update your submodules to their most recent upstream versions, use:
git submodule update --remote --merge
or
git submodule foreach git pull origin master

Create a meta-repo that holds scripts from various other repos

A. Use case:

Many applications allow user scripting. Usually the user's scripts are placed inside a specific directory that guarantees they can be accessed by the program. Since it is not possible to mix different git repos in one directory, manually updating them can be a chore. Even more so if the scripts are maintained separately/elsewhere.
This approach gives us the best of both worlds: we keep git's version control, and we keep all the scripts neatly in a single directory!
This means updating all of them in one go is as simple as running git submodule update --remote --merge.

B. How to:

  • We create a meta-scripts directory as meta-repo with a git subdirectory inside. This git subdirectory will hold the full repos from which we can then handpick the relevant scripts, and link them to the main meta-scripts directory:

    mkdir meta-scripts meta-scripts/git
    cd meta-scripts
    git init
    git submodule add https://github.com/user1/sub-project1.git git/sub-project1
    git submodule add https://github.com/user2/sub-project2.git git/sub-project2
    git commit -m 'Initial commit: add submodules'
    ln (-s) git/sub-project1/script1.lua
    ln (-s) git/sub-project2/script2.py
    git add script1.lua script2.py
    git commit -m 'Link scripts to main directory'
    git remote add origin https://github.com/my-user/meta-scripts.git
    git push -u origin master
    

    Note on the links:

    • If you use hardlinks (ln without the -s flag) the advantage is that the file contents of the links will be visible to git. The disadvantage is that git does not preserve the linking. When you clone the repo, the link between the files will no longer be present; both files will still be there, but without any relation between them.
    • In contrast, symbolic links (ln -s) do preserve their linking in git. However, the file contents of the links will not be visible to git; when you look at the symbolically linked file through git/github you will only see the path to the linked file (but the link will work on the filesystem).

    Regardlessly, symbolic links are more convenient since they won't break.

  • Then you can pull the meta-scripts repo wherever you need the scripts:

    cd /place/where/scripts/are/needed
    git clone https://github.com/my-user/meta-scripts.git .
    git submodule init
    git submodule update
    
  • To update all scripts/submodules in one go:

    git submodule update --remote --merge
    

$ git tag [tag-name]

List existing tags when no tag name is specified, or create a tag otherwise.
Note that there are 2 types of tags in git, which are created differently:

  • Lightweight tags:
    A lightweight tag is very much like a branch that doesn’t change; it's just a commit with a tag name.
    This is the type of tag that GitHub creates by default, when a Release is authored from the UI
    (although it will allow selecting an already existing annotated tag).

    => Create a lightweight tag with:

    $ git tag v1.0.0
    $ git show v1.0.0
    commit ca82a6dff817ec66f44342007202690a93763949 (HEAD -> master, tag: v1.0.0)
    Author: Scott Chacon <[email protected]>
    Date:   Mon Mar 17 21:52:11 2008 -0700
    
        Changed the version number
    
  • Annotated tags:
    An annotated tag is stored as a full object in the Git database; it is checksummed, contains the tagger name, email, date and tagging message. This is the type that npm version creates.

    Even though there is currently no way to create annotated tags from the Github UI, they're still recommended for the added information. This means you'll have to create them from the command line and push them to Github with e.g. git push --tags.

    => Create an annotated tag with:

    $ git tag -a v1.0.0 -m "Bump release to v1.0.0"
    (= git tag -am "Bump release to v1.0.0" v1.0.0) 
    $ git show v1.0.0
    tag v1.0.0
    Tagger: Ben Straub <[email protected]>
    Date:   Sat May 3 20:19:12 2014 -0700
    
    Bump release to v1.0.0
    
    commit ca82a6dff817ec66f44342007202690a93763949  (tag: v1.0.0)
    Author: Scott Chacon <[email protected]>
    Date:   Mon Mar 17 21:52:11 2008 -0700
    
        Changed the version number
    
⚠️ NOTE:
Tags are not automatically pushed to the remote, so it's recommended to explicitly push them if necessary:
git push --tags (origin master)
Or, if you only want to push a single tag:
git push origin v1.0.0
  • git tag -d v1.0.0
    Delete the v1.0.0 tag.
    To delete a remote tag, use:
    git push --delete origin v1.0.0

  • git tag -l
    List existing tags. Identical to git tag without arguments.  

Custom git commands

a) git aliases

It is possible to create new git commands by aliasing existing command(s) to a shorter word.
To do so, add them in ~/.gitconfig:

[alias]
    cm = commit -m              # zsh: gcmsg  (msg = message)
    cam = commit -am            # zsh: gcam
    difs = diff --staged        # zsh: gds
    logs = log --name-status
    prev = checkout HEAD^1
    next = "!sh -c 'git log --reverse --pretty=%H master |\
        awk \"/$(git rev-parse HEAD)/{getline;print}\" | xargs git checkout'"

    # Count the commits for the branch you are on:
    count = "!echo \"$(git rev-list --count HEAD) commits total on this branch\""

    # Count the commits for the branch you are on per user:
    countall = "!git shortlog -sn | cat; \
                 echo -n ' +  __________________________\n\n    '; \
                 echo \"$(git rev-list --count HEAD) commits total on this branch\""

    # Print essential info from both git log and git status:
    print = "!git status -s && git log -5 --oneline | cat"

    # Undo the last commit, but keep its changes staged, so recommitting is easy:
    undo = "!git reset HEAD~ --soft; \
             echo -n 'HEAD is now at '; \
             git log -1 --oneline | cat; \
             git status -s"

Even though we can put shell scripts inline inside .gitconfig, this is not advisable for anything but the most basic scripts.
By using the below method for separate git-subcommand scripts, we gain control and maintainability.

NOTE:
When using zshell with oh-my-zsh, you have access to the default zshell git aliases.
However, there are some unfortunate inconsistencies in these defaults, which you can rectify by overwriting them in ~/.zshrc after oh-my-zsh was loaded:

# Plugins to load:
plugins=(git)

# Load oh-my-zsh (omz):
source $ZSH/oh-my-zsh.sh

# Aliases overwriting some inconsistent omz git defaults:
alias gba='git branch -a | cat'
alias gcm='git commit -m'    # omz has gcm='git checkout master', but gco='git checkout'
alias gcom='git checkout master'
alias gcod='git checkout develop'
alias gcob='git checkout -b'
alias gl='git log --name-status'  # omz has gl='git pull', and glg='git log --stat'
alias gp='git status -s && git log -5 --oneline | cat' # git print
alias gph='git push'
alias gpl='git pull'
alias gs='git status'
alias gst='git stash'
alias gsta='git stash apply'
alias gsh='git show'

With these changes, the most useful shortcuts are:

Alias Command
g git
ga git add
gb git branch
gba git branch -a | cat
gbd git branch -d
gcam git commit -am
gcm git commit -m
gco git checkout
gcob git checkout -b
gcom git checkout master
gcp git cherry-pick
gd git diff
gds git diff --staged
gl git log --name-status
gp (= git print) git status -s && git log -5 --oneline | cat
gph git push
gpl git pull
gs git status
gsh git show
gst git stash

b) Separate git-subcommand scripts

  • Procedure:
    • Create a shell script named git-subcommand
    • Place it somewhere in your system's path, e.g. /usr/local/bin/git-subcommand
    • Give it executable permissions: sudo chmod +x /usr/local/bin/git-subcommand
    • It is now available in the shell by running git subcommand

$ git child

#!/bin/sh

# Checkout the child (newer) commit.
# Place this script somewhere in your path, e.g. /usr/local/bin/git-child

alias commits="git log --all --ancestry-path ^HEAD --format=format:%H"
children=$(commits)

# Failure case:
if [ $? != 0 ]; then exit 1
elif [ -z "$children" ]; then
    echo -n "You're at the tip of this branch\nHEAD remains at "
    git log -1 --oneline | cat
    exit 1
# Success case:
elif [ "$(commits | wc -l)" = 0 ]; then
    child=$(git branch --contains $children)
else
    child=$(commits | tail -n 1)
fi

git checkout $child
exit 0

$ git delete

#!/bin/sh

# Delete the last commit from both git as well as the working directory, and display status.
# Place this script somewhere in your path, e.g. /usr/local/bin/git-delete

delete=$(git reset HEAD~ --keep)
if [ $? != 0 ]; then
    # Failure case:
    echo -n "HEAD remains at "
    git log -1 --oneline | cat
    exit 1
else # Success case:
    echo -n "HEAD is now at "
    git log -1 --oneline | cat
    git status -s
    exit 0
fi