code-as-data

Introduction

Programming projects grow over time. New contributors join, new features are added, complexity increases. As this happens, the version-control history of the project becomes a valuable source of data. You might want to know:

• who is the main contributor to this particular file?

• if I change this file, are there any other files that I might need to look at?

• are there any files that should be combined together, or split apart?

Here, we have obtained the source code and version-control history for a range of R packages. For each package, the changes to individual files can be mapped out over history.

The packages studied were all related to package-development.

There is a shiny-app that presents the results from the analysis at shinyapps.io.

Environment

This project uses

• conda to manage the python-based and command-line tools and to install the base R release
• and renv to manage the installation of any R packages

To set up the project, first install the conda environment, activate it, then install the renv environment

Once you've got conda installed:

# To create an identical environment (req Ubuntu)
conda create --name code-as-data --file conda/requirements.txt

# To create a similar environment
conda env create --file conda/environment.yml --name code-as-data

Activate the conda environment

conda activate code-as-data

To install the R environment (and also make any non-version-controlled directories: data, results etc):

./setup

To run the analysis:

Once the environment is setup and activated use the bash script to run the workflow

./run

If you want the data or results to be stored to a specific location, set up links to these positions before running ./run.