Reading List

The most recent articles from a list of feeds I subscribe to.

How HEAD works in git

Hello! The other day I ran a Mastodon poll asking people how confident they were that they understood how HEAD works in Git. The results (out of 1700 votes) were a little surprising to me:

  • 10% “100%”
  • 36% “pretty confident”
  • 39% “somewhat confident?”
  • 15% “literally no idea”

I was surprised that people were so unconfident about their understanding – I’d been thinking of HEAD as a pretty straightforward topic.

Usually when people say that a topic is confusing when I think it’s not, the reason is that there’s actually some hidden complexity that I wasn’t considering. And after some follow up conversations, it turned out that HEAD actually was a bit more complicated than I’d appreciated!

Here’s a quick table of contents:

HEAD is actually a few different things

After talking to a bunch of different people about HEAD, I realized that HEAD actually has a few different closely related meanings:

  1. The file .git/HEAD
  2. HEAD as in git show HEAD (git calls this a “revision parameter”)
  3. All of the ways git uses HEAD in the output of various commands (<<<<<<<<<<HEAD, (HEAD -> main), detached HEAD state, On branch main, etc)

These are extremely closely related to each other, but I don’t think the relationship is totally obvious to folks who are starting out with git.

the file .git/HEAD

Git has a very important file called .git/HEAD. The way this file works is that it contains either:

  1. The name of a branch (like ref: refs/heads/main)
  2. A commit ID (like 96fa6899ea34697257e84865fefc56beb42d6390)

This file is what determines what your “current branch” is in Git. For example, when you run git status and see this:

$ git status
On branch main

it means that the file .git/HEAD contains ref: refs/heads/main.

If .git/HEAD contains a commit ID instead of a branch, git calls that “detached HEAD state”. We’ll get to that later.

(People will sometimes say that HEAD contains a name of a reference or a commit ID, but I’m pretty sure that that the reference has to be a branch. You can technically make .git/HEAD contain the name of a reference that isn’t a branch by manually editing .git/HEAD, but I don’t think you can do it with a regular git command. I’d be interested to know if there is a regular-git-command way to make .git/HEAD a non-branch reference though, and if so why you might want to do that!)

HEAD as in git show HEAD

It’s very common to use HEAD in git commands to refer to a commit ID, like:

  • git diff HEAD
  • git rebase -i HEAD^^^^
  • git diff main..HEAD
  • git reset --hard HEAD@{2}

All of these things (HEAD, HEAD^^^, HEAD@{2}) are called “revision parameters”. They’re documented in man gitrevisions, and Git will try to resolve them to a commit ID.

(I’ve honestly never actually heard the term “revision parameter” before, but that’s the term that’ll get you to the documentation for this concept)

HEAD in git show HEAD has a pretty simple meaning: it resolves to the current commit you have checked out! Git resolves HEAD in one of two ways:

  1. if .git/HEAD contains a branch name, it’ll be the latest commit on that branch (for example by reading it from .git/refs/heads/main)
  2. if .git/HEAD contains a commit ID, it’ll be that commit ID

next: all the output formats

Now we’ve talked about the file .git/HEAD, and the “revision parameter” HEAD, like in git show HEAD. We’re left with all of the various ways git uses HEAD in its output.

git status: “on branch main” or “HEAD detached”

When you run git status, the first line will always look like one of these two:

  1. on branch main. This means that .git/HEAD contains a branch.
  2. HEAD detached at 90c81c72. This means that .git/HEAD contains a commit ID.

I promised earlier I’d explain what “HEAD detached” means, so let’s do that now.

detached HEAD state

“HEAD is detached” or “detached HEAD state” mean that you have no current branch.

Having no current branch is a little dangerous because if you make new commits, those commits won’t be attached to any branch – they’ll be orphaned! Orphaned commits are a problem for 2 reasons:

  1. the commits are more difficult to find (you can’t run git log somebranch to find them)
  2. orphaned commits will eventually be deleted by git’s garbage collection

Personally I’m very careful about avoiding creating commits in detached HEAD state, though some people prefer to work that way. Getting out of detached HEAD state is pretty easy though, you can either:

  1. Go back to a branch (git checkout main)
  2. Create a new branch at that commit (git checkout -b newbranch)
  3. If you’re in detached HEAD state because you’re in the middle of a rebase, finish or abort the rebase (git rebase --abort)

Okay, back to other git commands which have HEAD in their output!

git log: (HEAD -> main)

When you run git log and look at the first line, you might see one of the following 3 things:

  1. commit 96fa6899ea (HEAD -> main)
  2. commit 96fa6899ea (HEAD, main)
  3. commit 96fa6899ea (HEAD)

It’s not totally obvious how to interpret these, so here’s the deal:

  • inside the (...), git lists every reference that points at that commit, for example (HEAD -> main, origin/main, origin/HEAD) means HEAD, main, origin/main, and origin/HEAD all point at that commit (either directly or indirectly)
  • HEAD -> main means that your current branch is main
  • If that line says HEAD, instead of HEAD ->, it means you’re in detached HEAD state (you have no current branch)

if we use these rules to explain the 3 examples above: the result is:

  1. commit 96fa6899ea (HEAD -> main) means:
    • .git/HEAD contains ref: refs/heads/main
    • .git/refs/heads/main contains 96fa6899ea
  2. commit 96fa6899ea (HEAD, main) means:
    • .git/HEAD contains 96fa6899ea (HEAD is “detached”)
    • .git/refs/heads/main also contains 96fa6899ea
  3. commit 96fa6899ea (HEAD) means:
    • .git/HEAD contains 96fa6899ea (HEAD is “detached”)
    • .git/refs/heads/main either contains a different commit ID or doesn’t exist

merge conflicts: <<<<<<< HEAD is just confusing

When you’re resolving a merge conflict, you might see something like this:

<<<<<<< HEAD
def parse(input):
    return input.split("\n")
=======
def parse(text):
    return text.split("\n\n")
>>>>>>> somebranch

I find HEAD in this context extremely confusing and I basically just ignore it. Here’s why.

  • When you do a merge, HEAD in the merge conflict is the same as what HEAD was when you ran git merge. Simple.
  • When you do a rebase, HEAD in the merge conflict is something totally different: it’s the other commit that you’re rebasing on top of. So it’s totally different from what HEAD was when you ran git rebase. It’s like this because rebase works by first checking out the other commit and then repeatedly cherry-picking commits on top of it.

Similarly, the meaning of “ours” and “theirs” are flipped in a merge and rebase.

The fact that the meaning of HEAD changes depending on whether I’m doing a rebase or merge is really just too confusing for me and I find it much simpler to just ignore HEAD entirely and use another method to figure out which part of the code is which.

some thoughts on consistent terminology

I think HEAD would be more intuitive if git’s terminology around HEAD were a little more internally consistent.

For example, git talks about “detached HEAD state”, but never about “attached HEAD state” – git’s documentation never uses the term “attached” at all to refer to HEAD. And git talks about being “on” a branch, but never “not on” a branch.

So it’s very hard to guess that on branch main is actually the opposite of HEAD detached. How is the user supposed to guess that HEAD detached has anything to do with branches at all, or that “on branch main” has anything to do with HEAD?

that’s all!

If I think of other ways HEAD is used in Git (especially ways HEAD appears in Git’s output), I might add them to this post later.

If you find HEAD confusing, I hope this helps a bit!

Popular git config options

Hello! I always wish that command line tools came with data about how popular their various options are, like:

  • “basically nobody uses this one”
  • “80% of people use this, probably take a look”
  • “this one has 6 possible values but people only really use these 2 in practice”

So I asked about people’s favourite git config options on Mastodon:

what are your favourite git config options to set? Right now I only really have git config push.autosetupremote true and git config init.defaultBranch main set in my ~/.gitconfig, curious about what other people set

As usual I got a TON of great answers and learned about a bunch of very popular git config options that I’d never heard of.

I’m going to list the options, starting with (very roughly) the most popular ones. Here’s a table of contents:

All of the options are documented in man git-config, or this page.

pull.ff only or pull.rebase true

These two were the most popular. These both have similar goals: to avoid accidentally creating a merge commit when you run git pull on a branch where the upstream branch has diverged.

  • pull.rebase true is the equivalent of running git pull --rebase every time you git pull
  • pull.ff only is the equivalent of running git pull --ff-only every time you git pull

I’m pretty sure it doesn’t make sense to set both of them at once, since --ff-only overrides --rebase.

Personally I don’t use either of these since I prefer to decide how to handle that situation every time, and now git’s default behaviour when your branch has diverged from the upstream is to just throw an error and ask you what to do (very similar to what git pull --ff-only does).

merge.conflictstyle zdiff3

Next: making merge conflicts more readable! merge.conflictstyle zdiff3 and merge.conflictstyle diff3 were both super popular (“totally indispensable”).

The main idea is The consensus seemed to be “diff3 is great, and zdiff3 (which is newer) is even better!”.

So what’s the deal with diff3. Well, by default in git, merge conflicts look like this:

<<<<<<< HEAD
def parse(input):
    return input.split("\n")
=======
def parse(text):
    return text.split("\n\n")
>>>>>>> somebranch

I’m supposed to decide whether input.split("\n") or text.split("\n\n") is better. But how? What if I don’t remember whether \n or \n\n is right? Enter diff3!

Here’s what the same merge conflict look like with merge.conflictstyle diff3 set:

<<<<<<< HEAD
def parse(input):
    return input.split("\n")
||||||| b9447fc
def parse(input):
    return input.split("\n\n")
=======
def parse(text):
    return text.split("\n\n")
>>>>>>> somebranch

This has extra information: now the original version of the code is in the middle! So we can see that:

  • one side changed \n\n to \n
  • the other side renamed input to text

So presumably the correct merge conflict resolution is return text.split("\n"), since that combines the changes from both sides.

I haven’t used zdiff3, but a lot of people seem to think it’s better. The blog post Better Git Conflicts with zdiff3 talks more about it.

rebase.autosquash true

Autosquash was also a new feature to me. The goal is to make it easier to modify old commits.

Here’s how it works:

  • You have a commit that you would like to be combined with some commit that’s 3 commits ago, say add parsing code
  • You commit it with git commit --fixup OLD_COMMIT_ID, which gives the new commit the commit message fixup! add parsing code
  • Now, when you run git rebase --autosquash main, it will automatically combine all the fixup! commits with their targets

rebase.autosquash true means that --autosquash always gets passed automatically to git rebase.

rebase.autostash true

This automatically runs git stash before a git rebase and git stash pop after. It basically passes --autostash to git rebase.

Personally I’m a little scared of this since it potentially can result in merge conflicts after the rebase, but I guess that doesn’t come up very often for people since it seems like a really popular configuration option.

push.default simple, push.default current, push.autoSetupRemote true

These push options tell git push to automatically push the current branch to a remote branch with the same name.

  • push.default simple is the default in Git. It only works if your branch is already tracking a remote branch
  • push.default current is similar, but it’ll always push the local branch to a remote branch with the same name.
  • push.autoSetupRemote true is a little different – this one makes it so when you first push a branch, it’ll automatically set up tracking for it

I think I prefer push.autoSetupRemote true to push.default current because push.autoSetupRemote true also lets you pull from the matching remote branch (though you do need to push first to set up tracking). push.default current only lets you push.

I believe the only thing to be careful of with push.autoSetupRemote true and push.default current is that you need to be confident that you’re never going to accidentally make a local branch with the same name as an unrelated remote branch. Lots of people have branch naming conventions (like julia/my-change) that make this kind of conflict very unlikely, or just have few enough collaborators that branch name conflicts probably won’t happen.

init.defaultBranch main

Create a main branch instead of a master branch when creating a new repo.

commit.verbose true

This adds the whole commit diff in the text editor where you’re writing your commit message, to help you remember what you were doing.

rerere.enabled true

This enables rerere ("reuse recovered resolution"), which remembers how you resolved merge conflicts during a git rebase and automatically resolves conflicts for you when it can.

help.autocorrect 10

By default git’s autocorrect try to check for typos (like git ocmmit), but won’t actually run the corrected command.

If you want it to run the suggestion automatically, you can set help.autocorrect to 1 (run after 0.1 seconds), 10 (run after 1 second), immediate (run immediately), or prompt (run after prompting)

core.pager delta

The “pager” is what git uses to display the output of git diff, git log, git show, etc. People set it to:

  • delta (a fancy diff viewing tool with syntax highlighting)
  • less -x5,9 (sets tabstops, which I guess helps if you have a lot of files with tabs in them?)
  • less -F -X (not sure about this one, -F seems to disable the pager if everything fits on one screen if but my git seems to do that already anyway)
  • cat (to disable paging altogether)

I used to use delta but turned it off because somehow I messed up the colour scheme in my terminal and couldn’t figure out how to fix it. I think it’s a great tool though.

I believe delta also suggests that you set up interactive.diffFilter delta --color-only to syntax highlight code when you run git add -p.

diff.algorithm histogram

Git’s default diff algorithm often handles functions being reordered badly. For example look at this diff:

-.header {
+.footer {
     margin: 0;
 }

-.footer {
+.header {
     margin: 0;
+    color: green;
 }

I find it pretty confusing. But with diff.algorithm histogram, the diff looks like this instead, which I find much clearer:

-.header {
-    margin: 0;
-}
-
 .footer {
     margin: 0;
 }

+.header {
+    margin: 0;
+    color: green;
+}

Some folks also use patience, but histogram seems to be more popular. When to Use Each of the Git Diff Algorithms has more on this.

core.excludesfile: a global .gitignore

core.excludeFiles = ~/.gitignore lets you set a global gitignore file that applies to all repositories, for things like .idea or .DS_Store that you never want to commit to any repo. It defaults to ~/.config/git/ignore.

includeIf: separate git configs for personal and work

Lots of people said they use this to configure different email addresses for personal and work repositories. You can set it up something like this:

[includeIf "gitdir:~/code/<work>/"]
path = "~/code/<work>/.gitconfig"

url."git@github.com:".insteadOf 'https://github.com/'

I often accidentally clone the HTTP version of a repository instead of the SSH version and then have to manually go into ~/.git/config and edit the remote URL. This seems like a nice workaround: it’ll replace https://github.com in remotes with git@github.com:.

Here’s what it looks like in ~/.gitconfig since it’s kind of a mouthful:

[url "git@github.com:"]
	insteadOf = "https://github.com/"

One person said they use pushInsteadOf instead to only do the replacement for git push because they don’t want to have to unlock their SSH key when pulling a public repo.

A couple of other people mentioned setting insteadOf = "gh:" so they can git remote add gh:jvns/mysite to add a remote with less typing.

fsckobjects: avoid data corruption

A couple of people mentioned this one. Someone explained it as “detect data corruption eagerly. Rarely matters but has saved my entire team a couple times”.

transfer.fsckobjects = true
fetch.fsckobjects = true
receive.fsckObjects = true

submodule stuff

I’ve never understood anything about submodules but a couple of person said they like to set:

  • status.submoduleSummary true
  • diff.submodule log
  • submodule.recurse true

I won’t attempt to explain those but there’s an explanation on Mastodon by @unlambda here.

and more

Here’s everything else that was suggested by at least 2 people:

  • blame.ignoreRevsFile .git-blame-ignore-revs lets you specify a file with commits to ignore during git blame, so that giant renames don’t mess up your blames
  • branch.sort -committerdate, makes git branch sort by most recently used branches instead of alphabetical, to make it easier to find branches. tag.sort taggerdate is similar for tags.
  • color.ui false: to turn off colour
  • commit.cleanup scissors: so that you can write #include in a commit message without the # being treated as a comment and removed
  • core.autocrlf false: on Windows, to work well with folks using Unix
  • core.editor emacs: to use emacs (or another editor) to edit commit messages
  • credential.helper osxkeychain: use the Mac keychain for managing
  • diff.tool difftastic: use difftastic (or meld or nvimdiffs) to display diffs
  • diff.colorMoved default: uses different colours to highlight lines in diffs that have been “moved”
  • diff.colorMovedWS allow-indentation-change: with diff.colorMoved set, also ignores indentation changes
  • diff.context 10: include more context in diffs
  • fetch.prune true and fetch.prunetags - automatically delete remote tracking branches that have been deleted
  • gpg.format ssh: allow you to sign commits with SSH keys
  • log.date iso: display dates as 2023-05-25 13:54:51 instead of Thu May 25 13:54:51 2023
  • merge.keepbackup false, to get rid of the .orig files git creates during a merge conflict
  • merge.tool meld (or nvim, or nvimdiff) so that you can use git mergetool to help resolve merge conflicts
  • push.followtags true: push new tags along with commits being pushed
  • rebase.missingCommitsCheck error: don’t allow deleting commits during a rebase
  • rebase.updateRefs true: makes it much easier to rebase multiple stacked branches at a time. Here’s a blog post about it.

how to set these

I generally set git config options with git config --global NAME VALUE, for example git config --global diff.algorithm histogram. I usually set all of my options globally because it stresses me out to have different git behaviour in different repositories.

If I want to delete an option I’ll edit ~/.gitconfig manually, where they look like this:

[diff]
	algorithm = histogram

config changes I’ve made after writing this post

My git config is pretty minimal, I already had:

  • init.defaultBranch main
  • push.autoSetupRemote true
  • merge.tool meld
  • diff.colorMoved default (which actually doesn’t even work for me for some reason but I haven’t found the time to debug)

and I added these 3 after writing this blog post:

  • diff.algorithm histogram
  • branch.sort -committerdate
  • merge.conflictstyle zdiff3

I’d probably also set rebase.autosquash if making carefully crafted pull requests with multiple commits were a bigger part of my life right now.

I’ve learned to be cautious about setting new config options – it takes me a long time to get used to the new behaviour and if I change too many things at once I just get confused. branch.sort -committerdate is something I was already using anyway (through an alias), and I’m pretty sold that diff.algorithm histogram will make my diffs easier to read when I reorder functions.

that’s all!

I’m always amazed by how useful to just ask a lot of people what stuff they like and then list the most commonly mentioned ones, like with this list of new-ish command line tools I put together a couple of years ago. Having a list of 20 or 30 options to consider feels so much more efficient than combing through a list of all 600 or so git config options

It was a little confusing to summarize these because git’s default options have actually changed a lot of the years, so people occasionally have options set that were important 8 years ago but today are the default. Also a couple of the experimental options people were using have been removed and replaced with a different version.

I did my best to explain things accurately as of how git works right now in 2024 but I’ve definitely made mistakes in here somewhere, especially because I don’t use most of these options myself. Let me know on Mastodon if you see a mistake and I’ll try to fix it.

I might also ask people about aliases later, there were a bunch of great ones that I left out because this was already getting long.

Dealing with diverged git branches

Hello! One of the most common problems I see folks struggling with in Git is when a local branch (like main) and a remote branch (maybe also called main) have diverged.

There are two things that make this situation hard:

  • If you’re not used to interpreting git’s error messages, it’s nontrivial to even realize that your main has diverged from the remote main (git will often just give you an intimidating but generic error message like ! [rejected] main -> main (non-fast-forward) error: failed to push some refs to 'github.com:jvns/int-exposed')
  • Once you realize that your branch has diverged from the remote main, there no single clear way to handle it (what you need to do depends on the situation and your git workflow)

So let’s talk about a) how to recognize when you’re in a situation where a local branch and remote branch have diverged and b) what you can do about it! Here’s a quick table of contents:

Let’s start with what it means for 2 branches to have “diverged”.

what does “diverged” mean?

If you have a local main and a remote main, there are 4 basic configurations:

1: up to date. The local and remote main branches are in the exact same place. Something like this:

a - b - c - d
            ^ LOCAL
            ^ REMOTE

2: local is behind

Here you might want to git pull. Something like this:

a - b - c - d - e
    ^ LOCAL     ^ REMOTE

3: remote is behind

Here you might want to git push. Something like this:

a - b - c - d - e
    ^ REMOTE    ^ LOCAL

4: they’ve diverged :(

This is the situation we’re talking about in this blog post. It looks something like this:

a - b - c - d - e
        \       ^ LOCAL
         -- f 
            ^ REMOTE

There’s no one recipe for resolving this (how you want to handle it depends on the situation and your git workflow!) but let’s talk about how to recognize that you’re in that situation and some options for how to resolve it.

recognizing when branches are diverged

There are 3 main ways to tell that your branch has diverged.

way 1: git status

The easiest way to is to run git fetch and then git status. You’ll get a message something like this:

$ git fetch
$ git status
On branch main
Your branch and 'origin/main' have diverged, <-- here's the relevant line!
and have 1 and 2 different commits each, respectively.
  (use "git pull" to merge the remote branch into yours)

way 2: git push

When I run git push, sometimes I get an error like this:

$ git push
To github.com:jvns/int-exposed
 ! [rejected]        main -> main (non-fast-forward)
error: failed to push some refs to 'github.com:jvns/int-exposed'
hint: Updates were rejected because the tip of your current branch is behind
hint: its remote counterpart. Integrate the remote changes (e.g.
hint: 'git pull ...') before pushing again.
hint: See the 'Note about fast-forwards' in 'git push --help' for details.

This doesn’t always mean that my local main and the remote main have diverged (it could just mean that my main is behind), but for me it often means that. So if that happens I might run git fetch and git status to check.

way 3: git pull

If I git pull when my branches have diverged, I get this error message:

$ git pull
hint: You have divergent branches and need to specify how to reconcile them.
hint: You can do so by running one of the following commands sometime before
hint: your next pull:
hint:
hint:   git config pull.rebase false  # merge
hint:   git config pull.rebase true   # rebase
hint:   git config pull.ff only       # fast-forward only
hint:
hint: You can replace "git config" with "git config --global" to set a default
hint: preference for all repositories. You can also pass --rebase, --no-rebase,
hint: or --ff-only on the command line to override the configured default per
hint: invocation.
fatal: Need to specify how to reconcile divergent branches.

This is pretty clear about the issue (“you have divergent branches”).

git pull doesn’t always spit out this error message though when your branches have diverged: it depends on how you configure git. The three other options I’m aware of are:

  1. if you set git config pull.rebase false, it’ll automatically start merging the remote main
  2. if you set git config pull.rebase true, it’ll automatically start rebasing onto the remote main
  3. if you set git config pull.ff only, it’ll exit with the error fatal: Not possible to fast-forward, aborting.

Now that we’ve talked about some ways to recognize that you’re in a situation where your local branch has diverged from the remote one, let’s talk about what you can do about it.

there’s no one solution

There’s no “best” way to resolve branches that have diverged – it really depends on your workflow for git and why the situation is happening.

I use 3 main solutions, depending on the situation:

  1. I want to keep both sets of changes on main. To do this, I’ll run git pull --rebase.
  2. The remote changes are useless and I want to overwrite them. To do this, I’ll run git push --force
  3. The local changes are useless and I want to overwrite them. To do this, I’ll run git reset --hard origin/main

Here are some more details about all 3 of these solutions.

solution 1.1: git pull --rebase

This is what I do when I want to keep both sets of changes. It rebases main onto the remote main branch. I mostly use this in repositories where I’m doing all of my work on the main branch.

You can configure git config pull.rebase true, to do this automatically every time, but I don’t because sometimes I actually want to use solutions 2 or 3 (overwrite my local changes with the remote, or the reverse). I’d rather be warned “hey, these branches have diverged, how do you want to handle it?” and decide for myself if I want to rebase or not.

solution 1.2: git pull --no-rebase

This starts a merge between the local and remote main. Here you’ll need to:

  1. Run git pull --no-rebase. This starts a merge and (if it succeeds) opens a text editor so that you can confirm that you want to commit the merge
  2. Save the file in your text editor.

I don’t have too much to say about this because I’ve never done it. I always use rebase instead. That’s a personal workflow choice though, lots of people have very legitimate reasons to avoid rebase.

solution 2.1: git push --force

Sometimes I know that the work on the remote main is actually useless and I just want to overwrite it with whatever is on my local main.

I do this pretty often on private repositories where I’m the only committer, for example I might:

  • git push some commits
  • belatedly decide I want to change the most recent commit
  • make the changes and run git commit --amend
  • run git push --force

Of course, if the repository has many different committers, force-pushing in this way can cause a lot of problems. On shared repositories I’ll usually enable github branch protection so that it’s impossible to force push.

solution 2.2: git push --force-with-lease

I’ve still never actually used git push --force-with-lease, but I’ve seen a lot of people recommend it as an alternative to git push --force that makes sure that nobody else has changed the branch since the last time you pushed or fetched, so that you don’t accidentally blow their changes away.

Seems like a good option. I did notice that --force-with-lease isn’t foolproof though – for example this git commit talks about how if you use VSCode’s autofetching feature to continuously git fetch, then --force-with-lease won’t help you.

Apparently now Git also has --force-with-lease --force-if-includes (documented here), which I think checks the reflog to make sure that you’ve already integrated the remote branch into your branch somehow. I still don’t totally understand this but I found this stack overflow conversation helpful.

solution 3.1: git reset --hard origin/main

You can use this as the reverse of git push --force (since there’s no git pull --force). I do this when I know that my local work shouldn’t be there and I want to throw it away and replace it with whatever’s on the remote branch.

For example, I might do this if I accidentally made a commit to main that actually should have been on new branch. In that case I’ll also create a new branch (new-branch in this example) to store my local work on the main branch, so it’s not really being thrown away.

Fixing that problem looks like this:

git checkout main

# 1. create `new-branch` to store my work
git checkout -b new-branch   

# 2. go back to the `main` branch I messed up
git checkout main            

# 3. make sure that my `origin/main` is up to date
git fetch                    

# 4. double check to make sure I don't have any uncomitted 
# work because `git reset --hard` will blow it away                                       
git status                   

# 5. force my local branch to match the remote `main`                               
#    NOTE: replace `origin/main` with the actual name of the
#    remote/branch, you can get this from `git status`.
git reset --hard origin/main  

This “store your work on main on a new branch and then git reset --hard” pattern can also be useful if you’re not sure yet how to solve the conflict, since most people are more used to merging 2 local branches than dealing with merging a remote branch.

As always git reset --hard is a dangerous action and you can permanently lose your uncommitted work. I always run git status first to make sure I don’t have any uncommitted changes.

Some alternatives to using git reset --hard for this:

  • check out some other branch and run git branch -f main origin/main.
  • check out some other branch and run git fetch origin main:main --force

that’s all!

I’d never really thought about how confusing the git push and git pull error messages can be if you’re not used to reading them.

Inside .git

Hello! I posted a comic on Mastodon this week about what’s in the .git directory and someone requested a text version, so here it is. I added some extra notes too. First, here’s the image. It’s a ~15 word explanation of each part of your .git directory.

You can git clone https://github.com/jvns/inside-git if you want to run all these examples yourself.

Here’s a table of contents:

The first 5 parts (HEAD, branch, commit, tree, blobs) are the core of git.

HEAD: .git/head

HEAD is a tiny file that just contains the name of your current branch.

Example contents:

$ cat .git/HEAD
ref: refs/heads/main

HEAD can also be a commit ID, that’s called “detached HEAD state”.

branch: .git/refs/heads/main

A branch is stored as a tiny file that just contains 1 commit ID. It’s stored in a folder called refs/heads.

Example contents:

$ cat .git/refs/heads/main
1093da429f08e0e54cdc2b31526159e745d98ce0

commit: .git/objects/10/93da429...

A commit is a small file containing its parent(s), message, tree, and author.

Example contents:

$ git cat-file -p 1093da429f08e0e54cdc2b31526159e745d98ce0
tree 9f83ee7550919867e9219a75c23624c92ab5bd83
parent 33a0481b440426f0268c613d036b820bc064cdea
author Julia Evans <julia@example.com> 1706120622 -0500
committer Julia Evans <julia@example.com> 1706120622 -0500

add hello.py

These files are compressed, the best way to see objects is with git cat-file -p HASH.

tree: .git/objects/9f/83ee7550...

Trees are small files with directory listings. The files in it are called blobs.

Example contents:

$  git cat-file -p 9f83ee7550919867e9219a75c23624c92ab5bd83
100644 blob e69de29bb2d1d6434b8b29ae775ad8c2e48c5391	.gitignore
100644 blob 665c637a360874ce43bf74018768a96d2d4d219a	hello.py
040000 tree 24420a1530b1f4ec20ddb14c76df8c78c48f76a6	lib

The permissions here LOOK like unix permissions, but they’re actually super restricted, only 644 and 755 are allowed.

blobs: .git/objects/5a/475762c...

blobs are the files that contain your actual code

Example contents:

$ git cat-file -p 665c637a360874ce43bf74018768a96d2d4d219a	
print("hello world!")

Storing a new blob with every change can get big, so git gc periodically packs them for efficiency in .git/objects/pack.

reflog: .git/logs/refs/heads/main

The reflog stores the history of every branch, tag, and HEAD. For (mostly) every file in .git/refs, there’s a corresponding log in .git/logs/refs.

Example content for the main branch:

$ tail -n 1 .git/logs/refs/heads/main
33a0481b440426f0268c613d036b820bc064cdea
1093da429f08e0e54cdc2b31526159e745d98ce0
Julia Evans <julia@example.com>
1706119866 -0500
commit: add hello.py

each line of the reflog has:

  • before/after commit IDs
  • user
  • timestamp
  • log message

Normally it’s all one line, I just wrapped it for readability here.

remote-tracking branches: .git/refs/remotes/origin/main

Remote-tracking branches store the most recently seen commit ID for a remote branch

Example content:

$ cat .git/refs/remotes/origin/main
fcdeb177797e8ad8ad4c5381b97fc26bc8ddd5a2

When git status says “you’re up to date with origin/main”, it’s just looking at this. It’s often out of date, you can update it with git fetch origin main.

tags: .git/refs/tags/v1.0

A tag is a tiny file in .git/refs/tags containing a commit ID.

Example content:

$ cat .git/refs/tags/v1.0
1093da429f08e0e54cdc2b31526159e745d98ce0

Unlike branches, when you make new commits it doesn’t update the tag.

the stash: .git/refs/stash

The stash is a tiny file called .git/refs/stash. It contains the commit ID of a commit that’s created when you run git stash.

cat .git/refs/stash
62caf3d918112d54bcfa24f3c78a94c224283a78

The stash is a stack, and previous values are stored in .git/logs/refs/stash (the reflog for stash).

cat .git/logs/refs/stash
62caf3d9 e85c950f Julia Evans <julia@example.com> 1706290652 -0500	WIP on main: 1093da4 add hello.py
00000000 62caf3d9 Julia Evans <julia@example.com> 1706290668 -0500	WIP on main: 1093da4 add hello.py

Unlike branches and tags, if you git stash pop a commit from the stash, it’s deleted from the reflog so it’s almost impossible to find it again. The stash is the only reflog in git where things get deleted very soon after they’re added. (entries expire out of the branch reflogs too, but generally only after 90 days)

A note on refs:

At this point you’ve probably noticed that a lot of things (branches, remote-tracking branches, tags, and the stash) are commit IDs in .git/refs. They’re called “references” or “refs”. Every ref is a commit ID, but the different types of refs are treated VERY differently by git, so I find it useful to think about them separately even though they all use the same file format. For example, git deletes things from the stash reflog in a way that it won’t for branch or tag reflogs.

.git/config

.git/config is a config file for the repository. It’s where you configure your remotes.

Example content:

[remote "origin"] 
url = git@github.com: jvns/int-exposed 
fetch = +refs/heads/*: refs/remotes/origin/* 
[branch "main"] 
remote = origin 
merge refs/heads/main

git has local and global settings, the local settings are here and the global ones are in ~/.gitconfig hooks

hooks: .git/hooks/pre-commit

Hooks are optional scripts that you can set up to run (eg before a commit) to do anything you want.

Example content:

#!/bin/bash 
any-commands-you-want

(this obviously isn’t a real pre-commit hook)

the staging area: .git/index

The staging area stores files when you’re preparing to commit. This one is a binary file, unlike a lot of things in git which are essentially plain text files.

As far as I can tell the best way to look at the contents of the index is with git ls-files --stage:

$ git ls-files --stage
100644 e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 0	.gitignore
100644 665c637a360874ce43bf74018768a96d2d4d219a 0	hello.py
100644 e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 0	lib/empty.py

this isn’t exhaustive

There are some other things in .git like FETCH_HEAD, worktrees, and info. I only included the ones that I’ve found it useful to understand.

this isn’t meant to completely explain git

One of the most common pieces of advice I hear about git is “just learn how the .git directory is structured and then you’ll understand everything!”.

I love understanding the internals of things more than anyone, but there’s a LOT that “how the .git directory is structured” doesn’t explain, like:

  • how merges and rebases work and how they can go wrong (for instance this list of what can go wrong with rebase)
  • how exactly your colleagues are using git, and what guidelines you should be following to work with them successfully
  • how pushing/pulling code from other repositories works
  • how to handle merge conflicts

Hopefully this will be useful to some folks out there though.

some other references:

Do we think of git commits as diffs, snapshots, and/or histories?

Hello! I’ve been extremely slowly trying to figure how to explain every core concept in Git (commits! branches! remotes! the staging area!) and commits have been surprisingly tricky.

Understanding how git commits are implemented feels pretty straightforward to me (those are facts! I can look it up!), but it’s been much harder to figure out how other people think about commits. So like I’ve been doing a lot recently, I went on Mastodon and started asking some questions.

how do people think about Git commits?

I did a highly unscientific poll on Mastodon about how people think about Git commits: is it a snapshot? is it a diff? is it a list of every previous commit? (Of course it’s legitimate to think about it as all three, but I was curious about the primary way people think about Git commits). Here it is:

The results were:

  • 51% diff
  • 42% snapshot
  • 4% history of every previous commit
  • 3% “other”

I was really surprised that it was so evenly split between diffs and snapshots. People also made some interesting kind of contradictory statements like “in my mind a commit is a diff, but I think it’s actually implemented as a snapshot” and “in my mind a commit is a snapshot, but I think it’s actually implemented as a diff”. We’ll talk more about how a commit is actually implemented later in the post.

Before we go any further: when we say “a diff” or “a snapshot”, what does that mean?

what’s a diff?

What I mean by a diff is probably obvious: it’s what you get when you run git show COMMIT_ID. For example here’s a typo fix from rbspy:

diff --git a/src/ui/summary.rs b/src/ui/summary.rs
index 5c4ff9c..3ce9b3b 100644
--- a/src/ui/summary.rs
+++ b/src/ui/summary.rs
@@ -160,7 +160,7 @@ mod tests {
 ";

         let mut buf: Vec<u8> = Vec::new();
-        stats.write(&mut buf).expect("Callgrind write failed");
+        stats.write(&mut buf).expect("summary write failed");
         let actual = String::from_utf8(buf).expect("summary output not utf8");
         assert_eq!(actual, expected, "Unexpected summary output");
     }

You can see it on GitHub here: https://github.com/rbspy/rbspy/commit/24ad81d2439f9e63dd91cc1126ca1bb5d3a4da5b

what’s a snapshot?

When I say “a snapshot”, what I mean is “all the files that you get when you run git checkout COMMIT_ID”.

Git often calls the list of files for a commit a “tree” (as in “directory tree”), and you can see all of the files for the above example commit here on GitHub:

https://github.com/rbspy/rbspy/tree/24ad81d2439f9e63dd91cc1126ca1bb5d3a4da5b (it’s /tree/ instead of /commit/)

is “how Git implements it” really the right way to explain it?

Probably the most common piece of advice I hear related to learning Git is “just learn how Git represents things internally, and everything will make sense”. I obviously find this perspective extremely appealing (if you’ve spent any time reading this blog, you know I love thinking about how things are implemented internally).

But as a strategy for teaching Git, it hasn’t been as successful as I’d hoped! Often I’ve eagerly started explaining “okay, so git commits are snapshots with a pointer to their parent, and then a branch is a pointer to a commit, and…”, but the person I’m trying to help will tell me that they didn’t really find that explanation that useful at all and they still don’t get it. So I’ve been considering other options.

Let’s talk about the internals a bit anyway though.

how git represents commits internally: snapshots

Internally, git represents commits as snapshots (it stores the “tree” of the current version of every file). I wrote about this in In a git repository, where do your files live?, but here’s a very quick summary of what the internal format looks like.

Here’s how a commit is represented:

$ git cat-file -p 24ad81d2439f9e63dd91cc1126ca1bb5d3a4da5b
tree e197a79bef523842c91ee06fa19a51446975ec35
parent 26707359cdf0c2db66eb1216bf7ff00eac782f65
author Adam Jensen <adam@acj.sh> 1672104452 -0500
committer Adam Jensen <adam@acj.sh> 1672104890 -0500

Fix typo in expectation message

and here’s what we get when we look at this tree object: a list of every file / subdirectory in the repository’s root directory as of that commit:

$ git cat-file -p e197a79bef523842c91ee06fa19a51446975ec35
040000 tree 2fcc102acd27df8f24ddc3867b6756ac554b33ef	.cargo
040000 tree 7714769e97c483edb052ea14e7500735c04713eb	.github
100644 blob ebb410eb8266a8d6fbde8a9ffaf5db54a5fc979a	.gitignore
100644 blob fa1edfb73ce93054fe32d4eb35a5c4bee68c5bf5	ARCHITECTURE.md
100644 blob 9c1883ee31f4fa8b6546a7226754cfc84ada5726	CODE_OF_CONDUCT.md
100644 blob 9fac1017cb65883554f821914fac3fb713008a34	CONTRIBUTORS.md
100644 blob b009175dbcbc186fb8066344c0e899c3104f43e5	Cargo.lock
100644 blob 94b87cd2940697288e4f18530c5933f3110b405b	Cargo.toml

What this means is that checking out a Git commit is always fast: it’s just as easy for Git to check out a commit from yesterday as it is to check out a commit from 1 million commits ago. Git never has to replay 10000 diffs to figure out the current state or anything, because commits just aren’t stored as diffs.

snapshots are compressed using packfiles

I just said that Git commits are snapshots, but when someone says “I think of git commits as a snapshot, but I think internally they’re actually diffs”, that’s actually kind of true too! Git commits are not represented as diffs in the sense you’re probably used to (they’re not represented on disk as a diff from the previous commit), but the basic intuition that if you’re editing a 10,000 lines 500 times, it would be inefficient to store 500 copies of that file is right.

Git does have a way of storing files as differences from other ways. This is called “packfiles” and periodically git will do a garbage collection and compress your data into packfiles to save disk space. When you git clone a repository git will also compress the data.

I don’t have space for a full explanation of how packfiles work in this post (Aditya Mukerjee’s Unpacking Git packfiles is my favourite writeup of how they work). But here’s a quick summary of my understanding of how deltas work and how they’re different from diffs:

  • Objects are stored as a reference to an “original file”, plus a “delta”
  • the delta has a bunch of instructions like “read bytes 0 to 100, then insert bytes ‘hello there’, then read bytes 120 to 200”. It cobbles together bytes from the original plus new text. So there’s no notion of “deletions”, just copies and additions.
  • I think there are less layers of deltas: I don’t know how to actually check how many layers of deltas Git actually had to go through to get a given object, but my impression is that it usually isn’t very many. Probably less than 10? I’d love to know how to actually find this out though.
  • The “original file” isn’t necessarily from the previous commit, it could be anything. Maybe it could even be from a later commit? I’m not sure about that.
  • There’s no “right” algorithm for how to compute deltas, Git just has some approximate heuristics

what actually happens when you do a diff is kind of weird

When I run git show SOME_COMMIT to look at the diff for a commit, what actually happens is kind of counterintuitive. My understanding is:

  1. git looks in the packfiles and applies deltas to reconstruct the tree for that commit and for its parent.
  2. git diffs the two directory trees (the current commit’s tree, and the parent commit’s tree). Usually this is pretty fast because almost all of the files are exactly the same, so git can just compare the hashes of the identical files and do nothing almost all of the time.
  3. finally git shows me the diff

So it takes deltas, turns them into a snapshot, and then calculates a diff. It feels a little weird because it starts with a diff-like-thing and ends up with another diff-like-thing, but the deltas and diffs are actually totally different so it makes sense.

That said, the way I think of it is that git stores commits as snapshots and packfiles are just an implementation detail to save disk space and make clones faster. I’ve never actually needed to know how packfiles work for any practical reason, but it does help me understand how it’s possible for git commits to be snapshots without using way too much disk space.

a “wrong” mental model for git: commits are diffs

I think a pretty common “wrong” mental model for Git is:

  • commits are stored as diffs from the previous commit (plus a pointer to the parent commit(s) and an author and message).
  • to get the current state for a commit, Git starts at the beginning and replays all the previous commits

This model is obviously not true (in real life, commits are stored as snapshots, and diffs are calculated from those snapshots), but it seems very useful and coherent to me! It gets a little weird with merge commits, but maybe you just say it’s stored as a diff from the first parent of the merge.

I think wrong mental models are often extremely useful, and this one doesn’t seem very problematic to me for every day Git usage. I really like that it makes the thing that we deal with the most often (the diff) the most fundamental – it seems really intuitive to me.

I’ve also been thinking about other “wrong” mental models you can have about Git which seem pretty useful like:

  • commit messages can be edited (they can’t really, actually you make a copy of the commit with a new message, and the old commit continues to exist)
  • commits can be moved to have a different base (similarly, they’re copied)

I feel like there’s a whole very coherent “wrong” set of ideas you can have about git that are pretty well supported by Git’s UI and not very problematic most of the time. I think it can get messy when you want to undo a change or when something goes wrong though.

some advantages of “commit as diff”

Personally even though I know that in Git commits are snapshots, I probably think of them as diffs most of the time, because:

  • most of the time I’m concerned with the change I’m making – if I’m just changing 1 line of code, obviously I’m mostly thinking about just that 1 line of code and not the entire current state of the codebase
  • when you click on a Git commit on GitHub or use git show, you see the diff, so it’s just what I’m used to seeing
  • I use rebase a lot, which is all about replaying diffs

some advantages of “commit as snapshot”

I also think about commits as snapshots sometimes though, because:

  • git often gets confused about file moves: sometimes if I move a file and edit it, Git can’t recognize that it was moved and instead will show it as “deleted old.py, added new.py”. This is because git only stores snapshots, so when it says “moved old.py -> new.py”, it’s just guessing because the contents of old.py and new.py are similar.
  • it’s conceptually much easier to think about what git checkout COMMIT_ID is doing (the idea of replaying 10000 commits just feels stressful to me)
  • merge commits kind of make more sense to me as snapshots, because the merged commit can actually be literally anything (it’s just a new snapshot!). It helps me understand why you can make arbitrary changes when you’re resolving a merge conflict, and why it’s so important to be careful about conflict resolution.

some other ways to think about commits

Some folks in the Mastodon replies also mentioned:

  • “extra” out-of-band information about the commit, like an email or a GitHub pull request or just a conversation you had with a coworker
  • thinking about a diff as a “before state + after state”
  • and of course, that lots of people think of commits in lots of different ways depending on the situation

some other words people use to talk about commits might be less ambiguous:

  • “revision” (seems more like a snapshot)
  • “patch” (seems more like a diff)

that’s all for now!

It’s been very difficult for me to get a sense of what different mental models people have for git. It’s especially tricky because people get really into policing “wrong” mental models even though those “wrong” models are often really useful, so folks are reluctant to share their “wrong” ideas for fear of some Git Explainer coming out of the woodwork to explain to them why they’re Wrong. (these Git Explainers are often well-intentioned, but it still has a chilling effect either way)

But I’ve been learning a lot! I still don’t feel totally clear about how I want to talk about commits, but we’ll get there eventually.

Thanks to Marco Rogers, Marie Flanagan, and everyone on Mastodon for talking to me about git commits.