Reproduction scripts and hand-annotated data for the paper "Grading on a Curve: How Rust can Facilitate New Contributors while Decreasing Vulnerabilities"

Justin Tracey 8e55fda248 add link to paper in README.md 5 months ago
code afdd747eb6 inital grading-on-a-curve commit 5 months ago
data afdd747eb6 inital grading-on-a-curve commit 5 months ago
LICENSE ee0cf6bd41 add LICENSE file 5 months ago
README.md 8e55fda248 add link to paper in README.md 5 months ago

README.md

This repository contains the scripts and data necessary to reproduce the results of the "Grading on a Curve: How Rust can Facilitate New Contributors while Decreasing Vulnerabilities" paper. The data should reproduce by installing the requirements, cloning the repo, and running the reproduceResults.sh script in the code directory. Note that this will take some time, as it requires cloning seven additional git repositories (one of which is the very large Firefox git mirror), fetching data from the relatively slow Bugzilla API, and running gradient descent on a non-trivial amount of data. The exact amount of time will depend on your network connection and CPU, but running it overnight might be a good idea.

The software requirements for running the reproduceResults.sh script are as follows:

  • git
  • bash
  • jq
  • gradle
  • java and javac
  • python3 with the following modules:
    • numpy scipy pyplot urllib3

The reproduceResults.sh is intended to be fairly human-readable for a shell script, and its source can be used to find descriptions of what each step is doing and the role of each directory and file.