1 of 14

Version Control with Git

A brief Introduction�

John Robinson

2 of 14

What is Version Control?

Also known as:

Revision control

Source control

3 of 14

What is Version Control?

Software tools to help you manage your project:

Tracks changes to files

At points decided by you
With attribution

Any file can be tracked

Text ( csv, .py, .c, .r etc.) works best

4 of 14

Why use Version Control?

A more efficient use of storage

Reproducibility

This is an example of the type of situation we want to avoid - several reasons.

We’ve all been in this situation before

- multiple nearly-identical versions of the same file with no meaningful explanation of the differences are.

If we’re just dealing with Docs, some word processors let us deal with this a little better, like Microsoft Word (“Track Changes”) or Google Docs version history. BUT research isn’t just Words docs, it’s code and data and diagrams too.

Using version control means we don’t keep dozens of different versions of our files hanging about taking up space, and when we store a revision, we store an explanation of what changed.

So, if nothing else, using version control makes for a more space-efficient backup of your work.
When you use version control, at any point in the future, you can retrieve the correct versions of your documents, scripts or code. So, for example, a year after publication, you can get hold of the precise combination of scripts and data that you used to assemble a paper.

Version control makes reproducibility simpler. Without using version control it’s very hard to say that your research is truly reproducible

5 of 14

Why use Version Control?

3. Collaboration

6 of 14

How does VC work?

Version Control tracks changes to files:

Changes are tracked sequentially

7 of 14

How does VC work?

Different versions can be saved

8 of 14

How does VC work?

Multiple versions can be merged

9 of 14

Version Control Options

Subversion (svn) - Centralised
Mercurial (hg) - Distributed

Git (git) – Distributed

Most widely used in academia
N.B. GitHub != git

Git is the tool, GitHub is Co. + website. There are alternatives.

These are the most popular current Version Control systems:

Subversion has been around since about 2000. You may come across it you’re working with big, older projects, but we don’t recommend it as it has limitations:

With Subversion a single master copy of the repository (the files under version control) exists - the only place where all revision history is kept. If you’re working as team you need to connect to the central server to store a revision

Mercurial and Git are newer and work a little differently - they are Distributed Version control systems - each developer in a team has their own full copy of the repository which are then synchronised. You can use Git without a network connection and there’s no single point of failure.

Git was written by Linus Torvalds (who he?) (maintainer of the Linux kernel), to scratch his own itch, so if you think it’s idiosyncratic in places, you know who to blame.

10 of 14

Graphical Version Control

SourceTree

Git Kraken

Git GUI

11 of 14

Git commands (via Bash)

$ git config

Set global info

$ git init

Create a repository

$ git status

Show status

12 of 14

Git commands (via Bash)

$ git add

Track files

$ git commit

Log changes

$ git log

Show history

13 of 14

Optional further reading

Software Carpentry:

14 of 14

Optional further reading

Before you start, you’ll need the resources for that workshop, which are fetched (ironically, via git) in the Bash shell:

$ git clone https://github.com/Southampton-RSG/2019-03-13-southampton-swc.git