Git Introduction

1 Introduction

1.1 Learning Objectives

• Discuss benefits and challenges of version control systems (e.g., in the context of university study) and contrast decentralized ones with centralized ones
• Explain states of files under Git and apply commands to manage them
• Explain Feature Branch Workflow and apply it in sample scenarios
• Edit simple Markdown documents

1.2 Core Questions

• How to collaborate on shared documents as distributed team?

  “Magit screenshot” under CC0 1.0; from GitLab

  • Consider multiple people working on multiple files
    • Potentially in parallel on the same file
    • Think of group exercise sheet, project documentation, source code
• How to keep track of who changed what why?
• How to support unified/integrated end result?

1.3 Your Experiences?

• Briefly write down your own experiences.
  • Did you collaborate on documents
    • by sending them via e-mail,
    • by using shared (cloud) storage (e.g., Sciebo with OnlyOffice, Google),
    • by using collaborative editors (e.g., Sciebo with OnlyOffice, Etherpad, CodiMD, Overleaf)
    • by using version control systems (e.g., Git, SVN),
    • by using other means?
  • Why did you choose what alternative? What challenges arose? Do you bother to read Terms of Service when you entrust “your” documents and thoughts (each individual keystroke, including “deleted” parts) to third parties (e.g., in the cloud)?

1.4 Version Control Systems (VCSs)

• Synonyms: Version/source code/revision control system, source code management (VCS, SCM)
• Collaboration on repository of documents
  • Each document going through various versions/revisions
    • Each document improved by various authors
      • April 2012, Linux kernel 3.2: 1,316 developers from 226 companies

1.4.1 Major VCS features

• VCS keeps track of history

  • Who changed what why when?

    “Meeting arrows” under CC0 1.0; rotated from Pixabay

• VCS supports merging of versions into unified/integrated version
  • Integrate intermediate versions of single file with changes by multiple authors
• Copying of files is obsolete with VCSs
  • Do not create copies of files with names such as Git-Intro-Final-1.1.txt or Git-Intro-Final-reviewed-Alice.txt
    • Instead, use VCS mechanism, e.g., use tags with Git

2 Git Concepts

2.1 Git: A Decentralized VCS

• Various VCSs exist
  • E.g.: Git, BitKeeper, SVN, CVS
    • (Color code: decentralized, centralized)
• Git created by Linus Torvalds for the development of the kernel Linux

  • Reference: Pro Git book

    “Git Logo” by Jason Long under CC BY 3.0; from git-scm.com

  • Git as example of decentralized VCS
    • Every author has own copy of all documents and their history
    • Supports offline work without server connectivity
      • Of course, collaboration requires network connectivity
    • Distributed trust/control/visibility/surveillance

2.2 Key Terms: Fork, Commit, Push, Pull

• Fork/clone repository: Create copy of repository

  “Folder” under CC0 1.0; derived from Pixabay

  • Fork: Create copy within online Git platform
  • Clone: Create copy of remote repository on your machine

• Commit (aka check-in)

  “Folder” under CC0 1.0; derived from Pixabay

  • Make (some or all) changes permanent; announce them to version control system
  • Push: Publish (some or all) commits to remote repository
    • Requires authorization
  • Fetch (pull): Retrieve commits from remote repository (also merge them)

2.3 Key Terms: Branch, Merge

• Branches

  “Git Branches” by Atlassian under CC BY 2.5 Australia; dimension attributes added, from Atlassian

  • Alternative versions of documents, on which to commit
    • Without being disturbed by changes of others
    • Without disturbing others
      • You can share your branches if you like, though
• Merge
  • Combine changes of one branch into another branch
    • May or may not need to resolve conflicts
• (Don’t worry if this seems abstract, we’ll try this out.)

2.4 Git explained by Linus Torvalds

• Video at archive.org (Tech Talk, 2007, by Google Talks under CC BY-NC-SA 3.0)
  • Total length of 84 minutes, suggested viewing: 7:40 to 29:00

2.4.1 Review Questions

Prepare answers to the following questions

• What is the role of a VCS (or SCM, in Torvalds’ terminology)?
• What differences exist between decentralized and centralized VCSs?
  • By the way, Torvalds distinguishes centralized from distributed SCMs. I prefer “decentralized” over “distributed”. You?

3 Git Basics

3.1 In-Browser Tutorial

• Some students recommended this tutorial to try out Git commands in browser: https://learngitbranching.js.org/
  • Several levels of the tutorial cover Git commands that appear on later slides
    • Tab “Main”, Level “1: Introduction to Git Commits” introduces commit, branch, merge, rebase
    • Tab “Remote”, Level “1: Clone Intro” introduces clone, fetch, pull, push

3.2 Getting Started

• Quickstart for Git installation and setup for GitLab

• Really, please install Git now
• You may use Git without a server
  • Run git init in any directory
    • Keep track of your own files
  • By default, you work on the master branch
    • master is not more special than any other branch you may create

3.3 Git with Remote Repositories

• Download files from public repository: clone
  • git clone https://gitlab.com/oer/cs/programming.git
    • Later on, git pull merges changes to bring your copy up to date
• Contribute to remote repository
  • Two variants described later on

3.3.1 A quick check

3.4 First Steps with Git

• Prerequisites (see quickstart)
  • You installed Git
  • You performed the First-time Git setup

• Part 0
  • Create repository or clone one
    • git clone https://gitlab.com/oer/cs/programming.git
    • Creates directory programming
      • Change into that directory
      • Note presence of “real” contents and of sub-directory .git (with Git meta-data)

3.4.1 Part 1: Inspecting Status

• Execute git status
  • Output includes current branch (master) and potential changes
• Open some file in text editor and improve it
  • E.g., add something to Git-Introduction.org
• Create a new file, say, test.txt
• Execute git status again
  • Output indicates
    • Git-Introduction.org as not staged and modified
    • test.txt as untracked
    • Also, follow-up commands are suggested
      • git add to stage for commit
      • git checkout to discard changes

3.4.2 Part 2: Staging Changes

• Changes need to be staged before commit
  • git add is used for that purpose
  • Execute git add Git-Introduction.org
  • Execute git status
    • Output indicates Git-Introduction.org as to be committed and modified
• Modify Git-Introduction.org more
• Execute git status
  • Output indicates Git-Introduction.org as
    • To be committed and modified
      • Those are your changes added in Part 1
    • As well as not staged and modified
      • Those are your changes of Part 2

3.4.3 Part 3: Viewing Differences

• Execute git diff
  • Output shows changes that are not yet staged
    • Your changes of Part 2
• Execute git diff --cached
  • Output shows difference between staged changes and last committed version
• Execute git add Git-Introduction.org
• Execute both diff variants again
  • Lots of other variants exits
    • Execute git help diff
    • Similarly, help for other git commands is available

3.4.4 Part 4: Committing Changes

• Commit (to be committed) changes
  • Execute git commit -m "<what was improved>"
    • Where <what was improved> should be meaningful text
    • 50 characters or less
• Execute git status
  • Output no longer mentions Git-Introduction.org
    • Up to date from Git’s perspective
  • Output indicates that your branch advanced; git push suggested for follow-up
• Execute git log (press h for help, q to quit)
  • Output indicates commit history
  • Note your commit at top

3.4.5 Part 5: Undoing Changes

• Undo premature commit that only exists locally
  • Execute git reset HEAD~
    • (Don’t do this for commits that exist in remote places)
  • Execute git status and git log
    • Note that state before commit is restored
    • May apply more changes, commit later
• Undo git add with git reset
  • Execute git add Git-Introduction.org
  • Execute git reset Git-Introduction.org
• Restore committed version
  • Execute git checkout -- <file>
  • Warning: Local changes are lost

3.4.6 Part 6: Stashing Changes

• Save intermediate changes without commit
  • Execute git stash
    • If you performed git checkout ... on previous slide, change some file first
  • Execute git status and find yourself on previous commit
• Apply saved changes
  • Possibly on different branch or after git pull
  • Execute git stash apply
    • May lead to conflicts, to be resolved manually

3.4.7 Part 7: Branching

• Work on different branch
  • E.g., introduce new feature, fix bug
  • Execute git checkout -b testbranch
    • Option -b: Create new branch and switch to it
  • Execute git status and find yourself on new branch
    • With uncommitted modifications from master
    • Change more, commit on branch
    • Later on, merge or rebase with master
  • Execute git checkout master and git checkout testbranch to switch branches

3.4.8 Remotes (1)

• Show remote repositories, whose changes you track:
  • git remote -v
    • By default, remote after git clone is called origin
    • No remote exists after git init
    • For a forked project, one usually adds an upstream remote (see next two slides)
• Contribute to project, two variants
  1. Operation push (requires permission)
     • You can push to your own projects
     • E.g., push new branch to remote origin:
       • git push -u origin testbranch
  2. Use merge/pull requests for other projects (next slide)

3.4.9 Remotes (2)

• Contribute to some project, the upstream (section in Pro Git)
  • Projects follow different workflows; read contribution instructions first
  • E.g., Feature Branch Workflow
    • Fork upstream project (in GUI)
      • Which creates your own project
    • Clone it, full permissions on origin
    • Create branch for your improvements, commit, push branch to origin
    • In GUI, open merge request (GitLab) or pull request (GitHub) for branch
      • If accepted, its changes usually go into upstream’s master

3.4.10 Remotes (3)

• If your merge request was merged upstream, maybe update your fork to mirror upstream’ state
  • Goal: Update your master branch based on upstream’s master branch
  • Approach
    • Set up source project as remote upstream:
      • git remote add upstream <HTTPS-URL of source project>
    • Fetch and integrate state of upstream:
      • git fetch upstream
      • Integrate its master into your master, maybe with rebase:
        • git checkout master
        • git rebase upstream/master
    • Push updated master to your fork: git push

3.4.11 Review Questions

• As part of First Steps with Git, git status inspects repository, in particular file states
  • Recall that files may be untracked, if they are located inside a Git repository but not managed by Git
  • Other files may be called tracked
• Prepare answers to the following questions
  • Among the tracked files, which states can you identify from the demo? Which commands are presented to perform what state transitions?
  • Optional: Draw a diagram to visualize your findings

3.5 Merge vs Rebase

• Commands merge and rebase both unify two branches
• Illustrated subsequently
  • Same unified file contents in the end, but different views of history

3.5.1 Merge vs Rebase (1)

• Suppose you created branch for new feature and committed on that branch; in the meantime, somebody else committed to master

3.5.2 Merge vs Rebase (2)

• Merge creates new commit to combine both branches
  • Including all commits
  • Keeping parallel history

3.5.3 Merge vs Rebase (3)

• Rebase rewrites feature branch on master
  • Applies commits of feature on master
  • Cleaner end result, but branch’s history lost/changed

3.6 Git Workflows

• For collaboration, team needs to agree on git workflow
  • Several alternatives exist
• Feature Branch Workflow may be your starting point
  • Clone repository (maybe your forked version)
    • You need a repository to which you are allowed to push commits
  • Create separate branch for each independent contribution
    • E.g., bug fix, new feature, improved documentation
    • Enables independent work
  • Once done, push that branch, create pull/merge request, receive feedback
    • Pull/Merge request: special action asking maintainer to include your changes
    • Maintainer may merge branch into master

3.6.1 Sample Commands

git clone <project-URI>
# Then, later on retrieve latest changes:
git fetch origin
# See what to do, maybe pull when suggested in status output:
git status
git pull
# Create new branch for your work and switch to it:
git checkout -b nameForBranch
# Modify/add files, commit (potentially often):
git add newFile
git commit -m "Describe change"
# Push branch:
git push -u origin nameForBranch
# Ultimately, merge or rebase branch nameForBranch into branch master
git checkout master
git merge nameForBranch
# If conflict, resolve as instructed by git, commit.  Finally push:
git push

4 GitLab

4.1 GitLab Overview

• Web platform for Git repositories
  • https://about.gitlab.com/
  • Free software, which you could run on your own server
• Manage Git repositories
  • Web GUI for forks, commits, pull requests, issues, and much more
  • Notifications for lots of events
    • Not enabled by default
  • So-called Continuous Integration (CI) runners to be executed upon commit
    • Based on Docker images
    • Build and test your project (build executables, test them, deploy them, generate documentation, presentations, etc.)

4.2 GitLab in Action

• Exercise

5 Aside: Lightweight Markup Languages

5.1 Lightweight Markup

• Markup: “Tags” for annotation in text, e.g., indicate sections and headings, emphasis, quotations, …
• Lightweight markup
  • ASCII-only punctuation marks for “tags”
  • Human readable, simple syntax, standard text editor sufficient to read/write
  • Tool support
    • Comparison and merge, e.g., three-way merge
    • Conversion to target language (e.g. (X)HTML, PDF, EPUB, ODF)
      • Wikis, blogs
      • pandoc can convert between lots of languages

5.2 Markdown

• Markdown: A lightweight markup language
• Every Git repository should include a README file
  • What is the project about?
  • Typically, README.md in Markdown syntax
• Learning Markdown
  • In-browser tutorial (source code under MIT License)
  • Cheatsheet (under CC BY 3.0)

5.3 Org Mode

• Org mode: Another lightweight markup language
  • My favorite one
• For details see source file for this presentation as example

6 Conclusions

6.1 Summary

• VCSs enable collaboration on files
  • Source code, documentation, theses, presentations
• Decentralized VCSs such as Git enable distributed, in particular offline, work
  • Keeping track of files’ states
    • With support for subsequent merge of divergent versions
  • Workflows may prescribe use of branches for pull requests
• Documents with lightweight markup are particularly well-suited for Git management

6.2 Where to go from here?

• Version control is essential for DevOps
  • Combination of Development and Operations, see [JbA+16],[WFW+19]
  • Aiming for rapid software release cycles with high degree of automation and stability
• Variant based on Git is called GitOps, see [Lim18]
  • Self-service IT with proposals in pull requests (PRs)
  • Infrastructure as Code (IaC)

Bibliography

• [JbA+16] Jabbari, bin Ali, Petersen & Tanveer, What is DevOps? A Systematic Mapping Study on Definitions and Practices, in: Proceedings of the Scientific Workshop Proceedings of XP2016, 2016. https://doi.org/10.1145/2962695.2962707
• [Lim18] Limoncelli, GitOps: A Path to More Self-Service IT, Commun. ACM 61(9), 38-42 (2018). https://doi.org/10.1145/3233241
• [WFW+19] Wiedemann, Forsgren, Wiesche, Gewald & Krcmar, Research for Practice: The DevOps Phenomenon, Commun. ACM 62(8), 44-49 (2019). https://doi.org/10.1145/3331138

License Information

This document is part of a larger course. Source code and source files are available on GitLab under free licenses.