Determine metrics for measuring code quality

[missaugustina]

We can pull some of these from the Sonarqube tool, some might be available via the Github api, others we might need to do additional work on our own to calculate:

• code coverage (tons of interesting metrics listed in the sonarqube docs)
• count of tests - overall, by type (unit, functional)
• code complexity (ie, count of logical branches)
• count of lines of code
• count of comments
• count of lines of commented out code
• count of functions/methods
• count of classes
• count of files
• count of duplicated lines/blocks
• code performance profiling (harder, may not be possible and can't use Sonarqube for this one)

[missaugustina]

LOC/# files/# blanks/# comments - per language, per file. Allows snapshots in time. Distribution of commits across files (ie what files were changed most often based by commit history)

https://github.com/adamtornhill/code-maat

[missaugustina]

CLI utility for simple LOC based metrics - https://github.com/AlDanial/cloc

[missaugustina]

To clarify, there are two definitions of the word "quality". The value identified for this project refers to quality as a "degree of excellence". However, the metrics identified in this issue will only actually measure "distinctive characteristics". These metrics will only be useful as parameters to feed into a predictive model. If there is a high correlation between these metrics and repositories that use CI, then we can potentially predict what repos are likely to adopt CI in the near future (or repos that may be using CI but we can't easily identify it through other methods).

[missaugustina]

• https://insights.stackoverflow.com/survey/2016
• https://www.oreilly.com/ideas/code-quality-and-measuring-what-matters