---
tags: wp2, slides
title: RCA-Coverage testing
---

## Steps
1. Create a proof of concept, this tool https://github.com/giovannitangredi/sifis-test (only sifis algorithms) (end of February)
2. Add the other code coverage algorithms (SkunkScore and Crap) (Middle of March)
3. Search for simple repositories (sifis repositories for example), run the tool on it, vary the code of the repository a bit and then run the tool again. Analyze the differences and how the code evolves in this way we are able to analyze the algorithms. (later)
4. Integrate the tool inside grcov https://github.com/mozilla/grcov (later)
5. Evaluate if we can implement the analysis along the time, considering how a repository changes, using bugbug https://github.com/mozilla/bugbug (later and not sure)

## Intro

- Code coverage is good to catch bugs early
- A Full (100%) Coverage Testing may be too resource intensive to run and/or to write.
- With RCA we try to correlate code-complexity with odds-of-introducing-bugs
- We tried to use the code-complexity information to have a weighted code coverage
- Mozilla has already repositories in which the bug lifecycle (introduction to removal) is tracked

## Plan

We want to run the RCA-weighted code coverage over few selected repository and extract:

- The place of the bugs in the code-complexity source map
- How the code-coverage (weighted or not) evolved
- Try to estimate how many would be caught if the weight code coverage had been used instead of the plain one, assuming the flat coverage remained the same.

## Ingredients

- RCA
- grcov
- a repo with enough bugs tracked (~30?)
- A script to automate the data extraction