Creating Branches

Source code

Here's the code for the files we'll be using. Don't worry if you don't have Ruby installed; there's no need to run them. Our real goal is to learn how to track changes to these files.

decoder.rb

KEY = {
}

# Define a method that takes an array of numbers to decode.
def decode(numbers)
  text = ""
  # Process each number in the array.
  numbers.each do |number|
    # Get the letter that corresponds to this number.
    letter = KEY[number]
    # Add it onto the string.
    text += letter
  end
  # Return the decoded string.
  return text
end

try_decoder.rb

require './decoder'

puts decode([1, 3, 5])
puts decode([1, 18, 3, 1, 4, 5])

Git branches

• You've been working with one branch ever since you started using Git, perhaps without realizing it.
  • When you run git init, Git creates a new branch for you, called master. We can try this now: git init
  • If you run git status, it will show at the top that you're "On branch master".
• Now, branches act like alternate timelines for your file contents, but in reality, a branch is just a pointer to a particular commit. Don't worry if you don't understand that right now, we'll be circling back to that idea later.
  • So I can stage a file right now: git add decoder.rb
  • And I can make a commit: git commit -m 'Add decode function'
• If I run git log, you can see that commit:

commit e40647094c74d64fb7c4dee7351139c3ba644d9c (HEAD -> master)
Author: Jay McGavren <me@example.com>
Date:   Tue Sep 25 08:22:59 2018 -0700

    Add decode function

• See that "HEAD -> master" label on the right? That shows that the master branch is pointing to this commit.
  • HEAD is itself just a pointer to the branch that's currently checked out. We'll talk more about that later in the course.
• Right now our Git repository looks like this. There's only one commit.
• There's also only one branch, named master, and it's pointing to that commit.
• If I add another commit, the master branch will move to point to that commit.
  • I'll add another file: git add try_decoder.rb
  • And I'll make another commit: git commit -m 'Add main program'
  • If I show the log again, we'll see that the "master" label has moved: git log

commit 21428a2eee8e55becb6da879228446746ad6f1a8 (HEAD -> master)
Author: Jay McGavren <me@example.com>
Date:   Tue Sep 25 08:34:12 2018 -0700

    Add main program

commit e40647094c74d64fb7c4dee7351139c3ba644d9c
Author: Jay McGavren <me@example.com>
Date:   Tue Sep 25 08:22:59 2018 -0700

    Add decode function

• So now we've added a second commit, whose parent is the first commit.
• And the master branch has been moved to point to this second commit.

Making commits to the master branch is widely considered to be a bad idea, though. Most developers treat the master branch as code that has been thoroughly tested and is always ready to deploy to customers. If you've just finished some code that isn't thoroughly tested and ready to share, it shouldn't be committed to the master branch.

So where do we commit new code? To a new branch! Let's create one now.

• Suppose we wanted to add some unit tests.
• I'll create a new file, named test_decoder.rb.
• And I'll save this code to it.

test_decoder.rb

require 'minitest/autorun'
require './decoder'

class TestDecode < Minitest::Test
  def test_dad
    assert_equal("DAD", decode([4, 1, 4]))
  end
end

• I haven't committed this new file yet. When I do, I want it to be on a new branch.
• Commits we make on this new branch will be kept separately from commits we make on master or other branches.
  • We create a branch with the git branch subcommand. We type git branch followed by the name we want our new branch to have.
  • Branch names should be in all lower case. If there are multiple words, they should be separated by dashes: my-branch-name
  • But for this branch, we'll just go with a simpler name: git branch testing
• We can view the list of branches in the repo by typing git branch without a new branch name.
  • The active branch will have a star next to it in the output.
• If we make commits right now, they'll still go on the active branch, master. We need to switch to our testing branch so that commits go there instead.
  • We do that with the git checkout subcommand, followed by the name of the branch we want: git checkout testing
  • Let me run git log again: git log
  • You'll see the commits are the same, but there's now a new testing branch listed by the most recent commit.
  • Both the testing and master branches are pointing to this commit.
  • Also notice that the HEAD arrow is now pointing to testing instead of master.
  • We'll be talking about the HEAD reference more later, but basically this just means that testing is now the checked out branch instead of master.
  • The commits from the master branch are listed here because the testing branch is based on the master branch.

commit 21428a2eee8e55becb6da879228446746ad6f1a8 (HEAD -> testing, master)
Author: Jay McGavren <me@example.com>
Date:   Tue Sep 25 08:34:12 2018 -0700

    Add main program

commit e40647094c74d64fb7c4dee7351139c3ba644d9c
Author: Jay McGavren <me@example.com>
Date:   Tue Sep 25 08:22:59 2018 -0700

    Add decode function

• The repository looks like this.
• We've added a second branch, but there aren't any commits on it yet.
• The new branch is pointing to the same commit as the master branch.

• Now, let's make a commit on this testing branch.
  • I'll stage my file with the unit tests: git add test_decoder.rb
  • And commit it: git commit -m "Add unit test"
• If I run git log now, you'll see that there's a new commit, and that the testing branch pointer has been moved to point there.
  • The master branch, however, is still pointing to the same commit it was before.

commit 3c64f2dea660694d5da1ba7d5d1578f62d7919c8 (HEAD -> testing)
Author: Jay McGavren <me@example.com>
Date:   Tue Sep 25 18:25:05 2018 -0700

    Add unit test

commit 21428a2eee8e55becb6da879228446746ad6f1a8 (master)
Author: Jay McGavren <me@example.com>
Date:   Tue Sep 25 08:34:12 2018 -0700

    Add main program

commit e40647094c74d64fb7c4dee7351139c3ba644d9c
Author: Jay McGavren <me@example.com>
Date:   Tue Sep 25 08:22:59 2018 -0700

    Add decode function

• Here's what the repository looks like overall.
• A commit has been added, and the testing branch has been moved to point to that commit.
• The master branch is still pointing to the same commit.

• When you switch back and forth between branches, the working directory is changed to reflect the current branch.
  • Files that don't exist on this branch are removed.
  • Contents of each file are updated to reflect their status as of the last commit on this branch.
• The commit history is also updated to reflect the current branch.

Now let's add another commit to the testing branch.

• I have code for a second test, which I'll add to the test_decoder.rb file.

  def test_bead
    assert_equal("BEAD", decode([2, 5, 1, 4]))
  end

• Let me save and close that file.
• Now I'll commit it:
  • git add test_decoder.rb
  • git commit -m "Add test with more letters"
• Now that we have a second commit on our testing branch, let's try running git log again: git log
  • You'll see that the testing branch reference has moved to this latest commit.
  • The master branch, meanwhile, is still back at the same commit it was when we originally switched to a different branch.

commit 01c21ced52c664acf1362fa003ffc47bcde838ce (HEAD -> testing)
Author: Jay McGavren <me@example.com>
Date:   Wed Sep 26 20:00:18 2018 -0700

    Add test with more letters

commit 7bec24095bd1604e647912e6a8d4a0ff8061c129
Author: Jay McGavren <me@example.com>
Date:   Wed Sep 26 16:24:57 2018 -0700

    Add unit test

commit 21428a2eee8e55becb6da879228446746ad6f1a8 (master)
Author: Jay McGavren <me@example.com>
Date:   Tue Sep 25 08:34:12 2018 -0700

    Add main program

commit e40647094c74d64fb7c4dee7351139c3ba644d9c
Author: Jay McGavren <me@example.com>
Date:   Tue Sep 25 08:22:59 2018 -0700

    Add decode function

• The repository looks like this.
• The testing branch has been updated to point to the latest commit, but the master branch is still unmoved.

• The testing branch is local. It exists only on your computer.
  • It hasn't been sent over the network to any remote repos, and no network connection was needed to create it.
  • It is possible to push the branch to a remote repo. We'll be discussing that in a later stage of this course.

So now you have a new branch to hold your testing code. But what happens if you need to add code that isn't related to testing? We'll try creating another branch in the next video.