A short introduction to version control with git

Folder name: gitintro

One-time extra-special due date: you can have until Saturday 9/17 for this one

You will work with one or two partners assigned to you for this assignment. Do the individual work on your own computers, but share a single git repository for your project. Put all collaborators' names in the comments at the top of each source file.

Goals

To learn about:

Summary of your tasks

Here's the quick summary of your tasks. Much more detailed instructions are in the numbered sections below.

0. Detailed instructions (featuring Alice and Bob)

The rest of this document contains lots and lots of details. Go slowly, write down your questions as you go, and pay attention to error messages and other information that appears on your screens. Also, talk to your partner about each step: Did you understand what happened there? Should we try something different? Should we reread the instructions? What does that sentence mean? Maybe we should bug Jeff or post on Slack about this one. etc.

If you get stuck or just want a little elaboration on a concept, don't hesitate to ask for help, in class, in office hours, or on Slack.

For the rest of this document, and sometimes in other exercises, I will refer to the partner whose repo you're using for this project as Alice, and each other partner as Bob. This makes writing instructions a lot easier than referring to "the partner whose repo you're using for this project" and "the partner whose repo you're not using for this project". Alice and Bob have been used in computer-related exposition for a long time because they love communicating collaboratively and they're handy to have around.

Have fun!

1. Decide who's going to be Alice, and who is going to be Bob

If three of you are working together, pick Alice and then the other two people can be Bob1 and Bob2, and they can both do Bob's tasks.

2. Alice & Bob: look at Alice's CS257 repository on GitHub

Go to GitHub, login, and click on your cs257 repository. At your repo's home page, you should see a display showing the files in the repository. Right now, that's probably just README.md, maybe LICENSE if you added a license, and .gitignore.

A repository is essentially a folder in which you can store whatever files and subfolders you want. What makes a repository more than a folder, however, is that the version control system (git, in this case) keeps track of all the changes that get made to the files in the repository. This saving of history makes it possible for you to retrieve old work, track down the time when a bug was introduced (and who introduced it), collaborate on a project with a team of programmers, easily back up your work, etc. Version control systems are a bit of a hassle to learn for the first time, but they're indispensable programming tools and unquestionably worth the effort.

3. Alice: grab a local clone of your repository

Go to your repository's page on GitHub. There should be a big green button labeled "Code". Click it, and copy the URL that appears in the resulting popup window. You might be able to choose between a coupl URLs at this point; pick the one that starts with https.

Right now, your repository is stored only on the GitHub server. To add files to it, you'll need to clone a copy of the repository onto your working computer. You will then add files to your local copy of the repository and eventually, you'll manually push your changes back up to GitHub.

Create your clone like so:

You should now have a folder called "cs257" (or whatever you named your repository). This folder should contain README.md, maybe LICENSE, .gitignore, and a subfolder called .git. You can move cs257 wherever it's convenient for you to have it. (Curious about the contents of the .git folder? Go ahead and poke around to see what's there. It gets more interesting, of course, after you add some files.)

IMPORTANT FOR WINDOWS USERS: When you do this, you will need to be careful about where you put your files. If you put your repository inside the Linux file tree (by, say, doing "cd" at the terminal prompt to get to your Linux home directory and then doing "git clone ..." to put your git repository there), you'll run into file permissions trouble if you try to access your files using a Graphical User Interface (GUI) tool like VS Code or Brackets or Sublime or Notepad. There's a simple solution. You can use your terminal to access files in the normal Windows file system (say, on your C drive) by doing this:

cd /mnt/c/Users/YOUR_WINDOWS_USER_NAME/Desktop git clone https://github.com/...
which will put your repository on the Windows desktop. Of course, you can modify "Desktop" in this example to put the repo somewhere more convenient if you'd like, "Documents" or "Documents/cs257/" or whatever. By using the /mnt/c file tree, you enable the WSL terminal to work effectively with the normal Windows file permissions system.

4. Alice: add Bob as a collaborator on your repo

5. Bob: get a clone of Alice's repo

Once Bob is a collaborator on Alice's repo, Bob should git clone it to their working machine. You can add an extra argument to the clone command to name the new repo in a way that you'll remember it's Alice's repo:

git clone https://github.com/alice/cs257 cs257-alice

6. Alice: add a little bit of code

7. Bob: can you see Alice's changes?

(Alice? Hands off the keyboard. It's Bob's turn.)

Bob, at your terminal on your computer, cd to your clone of Alice's repository, and do

git pull

Does the hello.py file appear in your copy of the repo? It should. If it doesn't, try to figure out what went wrong, then ask for help if you're stuck.

8. Bob: make a change; Alice: get Bob's changes

Bob: Change something in hello.py (add a comment, add a print statement, whatever). Then do the git status/add/status/commit/push sequence from Alice's section above to get your changes included in the central copy on github.com. (Note that Bob will use Bob's PAT, not Alice's PAT in place of a push password, and because Alice added Bob as a collaborator on this repository, that will be good enough. Neither Alice nor Bob needs to know each other's PAT.)

Alice: do a "git pull" and see if you ended up with Bob's version of the code.

9. Bob: edit the .gitignore file

While we work on a project, it's common for the project directory to accumulate files that aren't really part of the project. For example, if you write program.py and module.py and program.py includes an "import module" statement, you may discover a module.pyc or a folder called __pycache__ in your working directory. This isn't source code, and it doesn't matter if you delete it—it'll just get recreated next time you "import module". So you don't want that litter included in your repository.

There are lots of other examples of files that shouldn't be saved in a repo. Like the .DS_Store files the macOS Finder generates whenever you look at a folder in a window or the .class files you get when you compile a .java file. Accidentally adding these files to your repository is all too easy. And since these files typically change every time you run the Python program or recompile the Java program or look at the folder in a window, all those changes get added to the repository every time you commit, which makes every clone of your repo take up a lot of unnecessary disk space. It's a mess, and we hate messes.

To prevent you from accidentally adding these extraneous files to your repository, you can tell git to ignore them completely via a .gitignore file. If Alice followed the setup instructions, she already has a .gitignore file full of weird stuff. Note that one line in .gitignore says "*.py[cod]" and another says "__pycache__/", which will cause git to ignore __pycache__ folders and .pyc, .pyo, and .pyd files.

To save macOS users a little pain, Bob is going to add the line ".DS_Store" to the .gitignore. .DS_Store files get created whenever you open a folder in the macOS Finder. You can delete them with no negative consequences, and you definitely do not want to include them in Alice's git repository.

IMPORTANT ODDITY: by default, macOS, Windows, and Unix command shells don't display files whose names start with a period. So if you open your repo's top-level folder in a macOS Finder window, you probably won't see your .gitignore file. Similarly, a simple:

ls

command at a Unix command line won't show you .gitignore. But this command does:

ls -a

(the "-a" stands for "all").

10. Alice: retrieve Bob's changes to .gitignore

What's the command again?

11. Alice & Bob: try git log

Execute git log on both of your machines. What do you see? What does it mean?

12. Alice & Bob: got a little time? Experiment.

Try just doing it all again, making small changes, adding files, etc., sending changes back and forth. Get comfortable with the basic edit/add/commit/pull/push workflow.

13. All done? Alice: delete Bob as a collaborator

It's good to keep your permissions as limited as possible. So even if you expect to work with Bob on the big project, you might want to delete them as a collaborator for the short term.

14. Yipee!

Congratualtions! You're using git.