Skip to content

Version control

Version control is a very important aspect of software development. In short, it allows to answer the following questions.

  • What was changed?
  • When was it changed?
  • Who made the change?
  • Why was the change made?

Using a version control system you can compare versions of your code, and if necessary, revert to a previous version.

In scientific computing, source code is part of the experimental method: even a small change can affect the results and potentially the conclusions. Version control lets you experiment safely, compare implementations, restore a known-working version, and identify the exact code version used to produce a result. Recording that version with the results provides an important link in the chain of scientific provenance.

Version control alone does not make a computation reproducible: the input data, parameters, software environment, and workflow must also be recorded. However, it makes the evolution of the source code traceable.

A short version-control demonstration illustrates these benefits by introducing a bug into a small scientific program, inspecting the change, and recovering the known-working version.

This is the traceable changes stage of the running example. The demonstration uses version control to answer a scientific question—why did the result change—without teaching Git mechanics in this course.

It is good practice to host your repositories on a service such as GitHub, GitLab or a hosting service provided by your organization. These environments all facilitate collaboration on software projects and make it easy to work in teams.

Participants who want hands-on practice with Git commands and hosting services can continue with the dedicated Version control with Git training.