Version control
with Git and GitHub

Carlos Matos // ISPUP // November 2023

What is ?

• Application that runs on your computer
• Helps you manage work done on projects
• It’s like the track changes option on Word, but on steroids
• Rigorous, powerful, scales up to multiple files (data, figures, reports, code)

Why use version control for data science?

Why use version control for data science?

• Organizing and recording your work becomes part of your workflow, instead of a separate and burdensome task
• Collaboration is more structured, improved asynchronous work and managing versions
• After the initial setup, almost effortless to put your project online is marginal
• Online exposure
• Works as a backup (safe, shareable and accessible from different places)

What is ?

• GitHub complements Git
  • Distribution mechanism for Git repositories
  • Git works locally, GitHub online
  • Control who has what type of access (view, edit, admin)
  • Easy to regularly reconcile files between contributors

GitHub features

• Markdown integration
• Issues lets users of your code report problems they found
• Pull requests by people that want to add features to your work
• Github Pages to quickly host static websites (like the one where these slides are hosted!)
• Easy to deploy packages or other functionality (e.g. Some R packages are only available on GitHub, not on CRAN

Git and GitHub jargon

• repo or repository - your project folder
• commit - save a snapshot to your repo
• hash - a computer generated id for each commit
• checkout - time travel to a specific commit
• branch - a label that points to a commit
• merge - combination of two or more branches
• remote - a repo that exists in a different location (e.g. GitHub, other PC)
• pull - get new commits from the remote to the repo
• push - send your new commits to the remote

Git and GitHub jargon

• repo or repository - your project folder

• commit - save a snapshot to your repo

• hash - a computer generated id for each commit
• checkout - time travel to a specific commit
• branch - a label that points to a commit
• merge - combination of two or more branches
• remote - a repo that exists in a different location (e.g. GitHub, other PC)
• pull - get new commits from the remote to the repo

• push - send your new commits to the remote

Git and GitHub workflow

• We will work almost exclusively with RStudio, not the command line!

How to revert to a previous commit?

gitGraph
    commit
    commit
    branch new_feature
    checkout new_feature
    commit
    commit
    checkout main
    merge new_feature
    commit id: "5-106eefa"
    commit

• Suppose that When you made commit #6, you made some mistakes and want to revert back to the state of your repo at moment #5.
• You would write in the terminal git reset 106eefa

Initial setup

1. Register a free account on GitHub
2. Install Git
3. Introduce yourself to git (username, email)

## install.packages("usethis")

library(usethis)
use_git_config(user.name = "Your Name", user.email = "email@example.org")

#configure the default branch name
usethis::git_default_branch_configure()

#Create token
#I highly recommend selecting “repo”, “user”, and “workflow” scopes (the default ones)
usethis::create_github_token()

Initial setup

4. Tell RStudio where to find Git (Tools > Global Options > Git/SVN should point to your Git install directory)
   1. usually something like “C:/Program Files/Git/bin/git.exe” on windows or “/usr/bin/git” on macOS
5. Create a project on GitHub. More detailed instructions available here
6. On the GitHub repository, click on Code and copy the HTTPS value
7. After that, create a new RProject
   1. File > New Project > Version Control > Git
8. If all went well, you should have a new folder with the same name as your GitHub repo
9. If you already had a project, move the contents to the repo directory

Folder structure

• a .git hidden folder exists at the project directory, that should not be deleted
• a .gitignore file will be present if you select R language option on GitHub when first creating your repo (recommended)

Exercises

• After moving the course scripts to the newly created project, stage the files, commit and push to GitHub.
• Make some changes and commit again. Check the results on git and GitHub.