Introduction to Git and GitHub
Michael A. Dolan, Ph.D
May 19, 2016
Outline
I. Introduction to source control
A. History and fundamental concepts behind source control
B. Centralized vs. distributed version control
II. Introduction to Git
A. What is Git? Basic Git concepts and architecture
B. Git workflows: Creating a new repo (adding, committing code)
C. HEAD
D. Git commands (checking out code)
E. Master vs branch concept
F. Creating a branch/switching between branches
G. Merging branches and resolving conflicts
III. Introduction to GitHub
A. What is GitHub? Basic GitHub concepts
B. GitHub in practice: Distributed version control
C. Cloning a remote repo
D. Fetching/Pushing to a remote repo
E. Collaborating using Git and GitHub
What is a ‘version control system?’
• a way to manage files and directories
• track changes over time
• recall previous versions
• ‘source control’ is a subset of a VCS.
Some history of source control…
(1972) Source Code Control System (SCCS)
- closed source, part of UNIX
(1982) Revision Control System(RCS)
- open source
(1986) Concurrent Versions System (CVS)
- open source
(2000) Apache Subversion (SVN)
- open source
…more history
(2000) BitKeeper SCM
- closed source, proprietary, used with
source code management of Linux kernel
- free until 2005
- distributed version control
Distributed version control
No central server
Every developer is a client, the server and the repository
Source: http://bit.ly/1SH4E23
What is git?
What is git?
• created by Linus Torvalds, April 2005
• replacement for BitKeeper to manage Linux kernel changes
• a command line version control program
• uses checksums to ensure data integrity
• distributed version control (like BitKeeper)
• cross-platform (including Windows!)
• open source, free
Popularity
https://www.openhub.net/repositories/compare
http://bit.ly/1QyLoOu
http://www.indeed.com/jobtrends/q-svn-q-git-q-subversion-q-
github.html?relative=1
Git distributed version control
• “If you’re not distributed, you’re not worth using.” – Linus Torvalds
• no need to connect to central server
• can work without internet connection
• no single failure point
• developers can work independently and merge their work later
• every copy of a Git repository can serve either as the server or as a client
(and has complete history!)
• Git tracks changes, not versions
• Bunch of little change sets floating around
Is Git for me?
• People primarily working with source code
• Anyone wanting to track edits (especially changes
to text files)
- review history of changes
- anyone wanting to share, merge changes
• Anyone not afraid of
command line tools
Most popular languages used with Git
• HTML
• CSS
• Javascript
• Python
• ASP
• Scala
Not as useful for image, movies,
• Shell scripts
• PHP music…and files that must be
• Ruby interpreted (.pdf, .psd, etc.)
• Ruby on Rails
• Perl
• Java
• C
• C++
• C#
• Objective C
• Haskell
• CoffeeScript
• ActionScript
How do I get it?
http://git-scm.com
Git install tip
• Much better to set up on a per-user basis
(instead of a global, system-wide install)
What is a repository?
• “repo” = repository
• usually used to organize a single project
• repos can contain folders and files, images,
videos, spreadsheets, and data sets – anything
your project needs
Two-tree architecture
other VCSs
Repository
checkout commit
working
Git uses a three-tree architecture
Repository
commit
checkout Staging index
add
working
A simple Git workflow
1. Initialize a new project in a directory:
git init
2. Add a file using a text editor to the directory
3. Add every change that has been made to the directory:
git add .
4. Commit the change to the repo:
git commit –m “important message here”
.
After initializing a new git repo…
Repository
3. Commit changes with
a message commit
Staging index
2. Add changes
add
1. Make changes
working
A note about commit messages
• Tell what it does (present tense)
• Single line summary followed by blank space
followed by more complete description
• Keep lines to <= 72 characters
• Ticket or bug number helps
Good and bad examples
Bad: “Typo fix”
Good: “Add missing / in CSS section”
Bad: “Updates the table. We’ll discuss next
Monday with Darrell.”
Bad: git commit -m "Fix login bug”
Good: git commit -m
How to I see what was done?
git log
“SHAs”
Checksums
generated by
SHA1
encryption
algorithm
The HEAD pointer
• points to a specific commit in repo
• as new commits are made, the pointer
changes
• HEAD always points to the “tip” of the
currently checked-out branch in the repo
• (not the working directory or staging index)
• last state of repo (what was checked out initially)
• HEAD points to parent of next commit (where writing the next commit
takes place)
Last commit
Parent of Parent of Parent of
master Y4f7uiPRRo… Pu87rRi4DD.. Qs2o0k64ja… i7Ewd37kL9…
HEAD
branch 9i5Tyh67dg.. oe48Hr3Gh9.. d3Ui94Hje4...
Which files were changed and where
do they sit in the three tree?
git status – allows one to see where files are in
the three tree scheme
Repository
commit
Staging
index
add
working
What changed in working directory?
git diff – compares changes to files between
repo and working directory
Line numbers in file Repository
Removed
Added commit
Staging
index
Note: git diff --staged - compares staging index to repo add
Note: git diff filename can be used as well working
Deleting files from the repo
git rm filename.txt
Repository
• moves deleted file change to staging commit
area Staging
index
add
• It is not enough to delete the file in
your working directory. You must working
commit the change.
Deleting files from the repo
Moving (renaming) files
git mv filename1.txt filename2.txt
Repository
commit
Note: File file1.txt was committed to repo earlier.
Staging
index
add
working
Good news!
git init
git status
git log
git add 75% of the time you’ll be using
git commit only these commands
git diff
git rm
git mv
What if I want to undo changes made
to working directory?
git checkout something
(where “something” is a file or an entire branch)
Repository
• git checkout will grab the
Staging
file from the repo checkout
index
• Example: git checkout -- file1.txt
working
(“checkout file ‘file1.txt’ from the current branch”)
What if I want to undo changes added
to staging area?
git reset HEAD filename.txt
Repository
commit
Staging
index
add
working
What if I want to undo changes
committed to the repo?
git commit --amend -m “message”
Repository
• allows one to amend a change to the last
commit commit
Staging
• anything in staging area will be amended index
to the last commit add
working
Note: To undo changes to older commits, make a new commit
HEAD
Parent of Parent of Parent of
master Y4f7uiPRRo… Pu87rRi4DD.. Qs2o0k64ja… i7Ewd37kL9…
Added ‘apple’ Added ‘plum’ Added ‘apple’
Obtain older versions
Repository
commit
Staging
index
add
working
git checkout 6e073c640928b -- filename.txt
Note: Checking out older commits places them into the
staging area
git checkout 6e073c640928b -- filename.txt
Repository
comit
Staging
index
working
Which files are in a repo?
git ls-tree tree-ish
tree-ish – a way to reference a repo
full SHA, part SHA, HEAD, others
blob = file, tree = directory
branching
• allows one to try new ideas
• If an idea doesn’t work, throw away the branch.
Don’t have to undo many changes to master
branch
• If it does work, merge ideas into master branch.
• There is only one working directory
Branching and merging example
SHA = a commit
HEAD
master
Y4f7uiPRRo Pu87rRi4DD Qs2o0k64ja i7Ewd37kL9 he8o9iKlreD kle987yYieo mN34i4uwQ
h4Rt5uEl9p Ge8r67elOp
new branch
HEAD changes from new
branch merged into
master
Source: http://hades.github.io/2010/01/git-your-friend-not-foe-vol-2-branches/
16 forks and 7 contributors to the master branch
red - commits pointed to by tags
blue - branch heads
white - merge and bifurcation commits.
Source: https://www.tablix.org/~avian/blog/archives/2014/06/vesna_drivers_git_visualization/
In which branch am I?
git branch
How do I create a new branch?
git branch new_branch_name
Note: At this point, both HEADs of the branches are pointing to the same
commit (that of master)
How do I switch to new branch?
git checkout new_branch_name
At this point, one can switch between branches, making commits, etc. in either
branch, while the two stay separate from one another.
Note: In order to switch to another branch, your current working directory
must be clean (no conflicts, resulting in data loss).
Comparing branches
git diff first_branch..second_branch
How do I merge a branch?
From the branch into which you want to merge another
branch….
git merge branch_to_merge
Note: Always have a clean working directory when merging
“fast-forward” merge occurs when HEAD of master
branch is seen when looking back
HEAD
master Y4f7uiPRRo Pu87rRi4DD Qs2o0k64ja i7Ewd37kL9
HEAD
new_branch h4Rt5uEl9p
“recursive” merge occurs by looking back and combining
ancestors to resolve merge
HEAD HEAD
Y4f7uiPRRo Pu87rRi4DD Qs2o0k64ja i7Ewd37kL9 he8o9iKlreD kle987yYieo mN34i4uwQ
HEAD
h4Rt5uEl9p Ge8r67elOp
merge conflicts
What if there are two changes to same line in two
different commits?
file1.txt file1.txt
apple banana
master new_feature
Resolving merge conflicts
Git will notate the conflict in the files!
Solutions:
1. Abort the merge using git merge –abort
2. Manually fix the conflict
3. Use a merge tool (there are many out there)
Graphing merge history
git log --graph --oneline --all --decorate
Tips to reduce merge pain
• merge often
• keep commits small/focused
• bring changes occurring to master into your
branch frequently (“tracking”)
What is ?
GitHub
• a platform to host git code repositories
• http://github.com
• launched in 2008
• most popular Git host
• allows users to collaborate on projects from anywhere
• GitHub makes git social!
• Free to start
fork
GitHub Forked
master
branch
remote pull request
pull/ someone
server fetch else’s master
push
origin/master “branch”
push
merge merge
origin/master
references remote
repo server branch and
tries to stay in sync
checkout commit
Local
Stage
add
Working
directory
Important to remember
Sometimes developers choose to place repo on
GitHub as a centralized place where everyone
commits changes, but it doesn’t have to be on
GitHub
Source: http://bit.ly/1rvzjp9
Copying (cloning) files from remote
repo to local machine
git clone URL <new_dir_name>
fork
GitHub Forked
master
branch
remote pull request
pull/ someone
server fetch else’s master
push
origin/master “branch”
push
merge
clone
repo
checkout commit
Local
Stage
add
Working
directory
How do I link my local repo to a remote repo?
git remote add <alias> <URL>
Note: This just establishes a connection…no files are copied/moved
Note: Yes! You may have more than one remote linked to your local
directory!
Which remotes am I linked to?
git remote
Pushing to a remote repo
git push local_branch_alias branch_name
Fetching from a remote repo
git fetch remote_repo_name
Fetch in no way changes a your working dir or any
commits that you’ve made.
• Fetch before you work
• Fetch before you push
• Fetch often
git merge must be done to merge fetched changes into
local branch
Collaborating with Git
Collaborating with Git
Email sent along with link; collaborator has read/write access.
Everyone has read access
If you want write access and you haven’t been invited, “fork” the
repo
fork
GitHub Forked
master
branch
remote pull request
pull/ someone
server fetch else’s master
push
origin/master “branch”
push
merge merge
repo
checkout commit
Local
Stage
add
Working
directory
GitHub Gist
https://gist.github.com/
Good resources
• Git from Git: https://git-scm.com/book/en/v2
• A guided tour that walks through the fundamentals of Git:
https://githowto.com
• Linus Torvalds on Git:
https://www.youtube.com/watch?v=idLyobOhtO4
• Git tutorial from Atlassian:
https://www.atlassian.com/git/tutorials/
• A number of easy-to-understand guides by the GitHub folks
https://guides.github.com
git commit -a
• Allows one to add to staging index and
commit at the same time
• Grabs everything in working direcotry
• Files not tracked or being deleted are not
included
git log --oneline
• gets first line and checksum of all commits in
current branch
git diff g5iU0oPe7x
When using checksum of older commit, will
show you all changes compared to those in
your working directory
Renaming and deleting branches
git branch –m/--move old_name new_name
git branch –d branch_name
Note: Must not be in branch_name
Note: Must not have commits in branch_name unmerged in
branch from which you are deleting
git branch –D branch_name
Note: If you are *really* sure that you want to delete branch
with commits
Tagging
• Git has the ability to tag specific points in history
as being important, such as releases versions
(v.1.0, 2.0, …)
git tag
Tagging
Two types of tags:
lightweight – a pointer to a specific comment –
basically a SHA stored in a file
git tag tag_name
annotated – a full object stored in the Git database –
SHA, tagger name, email, date, message
and can be signed and verified with GNU
Privacy Guard (GPG)
git tag –a tag_name –m “message”
How do I see tags?
git show tag_name
Lightweight tag
Annotated tag
