Besides the "basic" commands of Git, everyone has their own little Git tricks they use. I wanted to quickly write a list of my own which I tend to alias in my
.gitconfig. Scroll to the bottom to see some fun
git related commands that run outside of git! :)
I often forget to commit a file, or leave a
console.log in. I absolutely hate doing commits like
removed console.log. So instead, I add the file as if I was going to make a commit and run:
git commit --amend --reuse-message HEAD
Which will add the file to the last commit and reuse the old commit message. I alias this one as
git amend for quickfixes
NOTE Based on feedback below, it's also possible to do
git commit --amend --no-edit for the same effect.
Older branches often fall behind pretty far, so far that I have to get up to speed to eliminate build errors, ci errors, or just resolve conflicts. My favorite is to do the following:
git fetch origin # fetch latest origin git rebase origin/master
This way, I'm stacking my current branch commits on top of the latest version of master!
git log gets overwhelming. Due to my frequent use of the aforementioned
amend command, i tend to want to view just the last commit in my git log:
git log -1
Occasionally, I screw up a file unrelated to my branch. Mostly, that happens with lock files (mix.lock, package-lock.json, etc.). Rather than reverting a commit which probably contained a bunch of other stuff, I just "reset" the file back to an older version
git checkout hash-goes-here mix.lock
And then I can commit the fix!
An underrated command that I occasionally use. When a branch gets stale, it's sometimes easier to just get the stuff you really need from it rather than try to get the entire branch up to speed. A good example, for me, have been branches that involve UI/backend code that is no longer necessary. In that case, I might want to cherry pick only certain commits from the branch
git cherry-pick hash-goes-here
This will magically bring that commit over to the branch you're on. You can also do a list!
git cherry-pick first-hash second-hash third-hash
You can also do a range
git cherry-pick first-hash..last-hash
This is such a power-user feature that I rarely use it. I mean, once a year! But it's good to know about it. Sometimes, I lose commits. I delete a branch or reset or amend a commit I didn't mean to mess up.
In those situations, it's good to know
reflog exists. It's not a log of individual commits for the branch you're on, it's a log of all of your commits -- even ones that were on dead branches. However, the log gets emptied from time to time (pruned) so that only relevant information stays.
The command returns a log and what's useful is cherry-picking or rebasing on top of a commit. Very powerful when you pipe into
Aside from git commands, I like to also use some fun bash aliases to help my workflow
To get the name of the current branch, I have this alias:
alias git-branch="git branch | sed -n -e 's/^\* \(.*\)/\1/p'"
When I run
git-branch or run
$(git-branch) in another command, I'll get the name of the current branch I'm on.
NOTE Based on feedback in the comments, I switched this over to
git symbolic-ref --short HEAD which works just as well but you can actually read it.
While I'm sure this is doable in the
.gitconfig, I've yet to figure out how. When I run the first push on a new branch, I always get asked to setup the branch for upstream tracking. Here's my alias for that:
alias git-up="git branch | sed -n -e 's/^\* \(.*\)/\1/p' | xargs git push -u origin "
Now when I run
git-up, I push my current branch and setup upstream tracking!
Based on some of the very helpful feedback in the comments, I made some adjustments to what I'm using.
It looks like there are a bunch of new ways to get the current branch name. If you scroll up, you'll see that I've used a crazy
sed parsing command to get the branch name.
Here's my new alternative:
alias git-branch="git symbolic-ref --short HEAD"
And it seems to work exactly as you'd expect!