Skip to content

Latest commit

 

History

History
236 lines (155 loc) · 5.91 KB

remotes.md

File metadata and controls

236 lines (155 loc) · 5.91 KB
Raw

Git - Remotes

You should read about branches before starting this section.

So way back in the beginning we ran git init to create a new repository. However, as is often the case you're actually very likely to want to get a copy of an existing repository.

Clone

If you've ever used Git before it's quite possible you've cloned a repository, which contained a number of pre-existing files, commits and branches.

Normally you'll find yourself cloning something from the web, and in particular Github. However, there's nothing special about Github, and in fact for the purposes of this guide let's clone the repository we've just been working in.

> cd ..

# Clone the current "test" repository to another folder called "local"
> git clone test test-clone

Cloning into 'test-clone'...
done.

This cloned repository looks a little different than our test repository.

> cd test-clone
> git log --graph --oneline --decorate master

* 41ce7fa (HEAD, origin/master, origin/HEAD, master) Third commit
* 672e562 (origin/hello) Second commit
* 406bb3b Initial commit

Previously we've seen HEAD and master, but what is origin?

Let's see that again in slow motion.

Remote

It's actually possible to clone a repository in a slightly different way, which may help understand a few things.

> cd ..
> mkdir local
> cd local
> git init

Initialized empty Git repository in local/.git/

# This assumes you called the other git repository "test"
> git remote add origin "../test"

At this point we've added one "remote" to Git. You can think of it like an "alias", or short-hand, for that filepath (which is normally going to be a long URL). The name "origin", much like "master", is just a default when you first clone. Because we've manually added the remote rather than clone we can choose what we like, but let's stick with "origin".

Can you add more remotes to that repository?

Fetch

However, you should notice there there's no files or branches in that repository. git remote doesn't actually do anything apart from update configuration.

> ls -F
> ls -F .git/refs

heads/
tags/

We need to actually download the Git repository. You can do this with the fetch command.

> git fetch origin

remote: Counting objects: 15, done.
remote: Compressing objects: 100% (12/12), done.
remote: Total 15 (delta 2), reused 15 (delta 2), pack-reused 0
Unpacking objects: 100% (15/15), done.
From ../test
 * [new branch]      hello      -> origin/hello
 * [new branch]      master     -> origin/master

What's extermely powerful about Git is that fetch will download all of the repository, not just the files in master. It should contain all of the history too for all of the branches.

[Advanced] Can you fetch using the path/URL (ie ../test) instead of the remote name? What does it do differently?

> ls
> git branch

> git log --graph --oneline --decorate master

fatal: ambiguous argument 'master': unknown revision or path not in the working tree.

Hmm, what gives? We still don't have any files/branches. Or do we?

> ls -F .git/refs

heads/
remotes/
tags/

> ls -F .git/refs/remotes

origin/

> ls -F .git/refs/remotes/origin

hello
master

> cat .git/refs/remotes/origin/master

41ce7faa5b8b2cf480d44c3b12f0545fed9683c3

So we don't have any refs/heads, but we do have refs/remotes/origin.

Can you run the log --graph for this ref/commit?

Whatever old man. Show me the files!

Checkout

WARNING This next section can be quite subtle but is crucial. This is one place where Git tries to make things "easy" which leads to confusion when it stops being nice and expects you to know what just happened. You have been warned.

So how do we checkout these remote branches?

What people normally type:

> git checkout master

Branch master set up to track remote branch master from origin.
Already on 'master'

> git log --graph --oneline --decorate master

* 41ce7fa (HEAD, origin/master, master) Third commit
* 672e562 (origin/hello) Second commit
* 406bb3b Initial commit

What really happened.

> git branch master origin/master

Branch master set up to track remote branch master from origin.

> git checkout master

Already on 'master'

So what's the problem? Typing just git checkout master looks much easier.

Let's see that again, but with a twist.

> git branch my_master origin/master

Branch my_master set up to track remote branch master from origin.

> git checkout my_master

> git log --graph --oneline --decorate my_master

* 41ce7fa (HEAD, origin/master, my_master) Third commit
* 672e562 (origin/hello) Second commit
* 406bb3b Initial commit

Keep in mind that this is the exact same thing as before. The only difference is that we've given our local branch a different name.

  • The first rule of Git remotes is that you cannot checkout remote branches.
  • The second rule of Git remotes is that you cannot checkout remote branches

As a rule I always recommend people never create a local branch called master to avoid confusion that is likely to ensue.

If your local branch is (also) called master, what happens if you run git log master? Which branch are you asking about?

What happens if you do try to checkout a remote branch (eg. git checkout origin/master)? How does that change the HEAD file? Is that what you would expect?

[Advanced] The HEAD files points to a ref, the remote branches are just refs. What happens if you manually change the HEAD file to point to refs/remote/origin/master? If you do, and then make a commit, what happens if you fetch after that ?

Next

That's probably enough for one section. Read about pull to see how to get more changes over time.