Benchmarking Brakeman

Brakeman is typically very consistent in its runtimes, which makes it easy to benchmark.

To avoid I/O inconsistencies, I recommend quiet mode and piping the report to /dev/null:

brakeman -q -o /dev/null

To avoid startup, library loading, and report generation penalties, I recommend using the scan time reported by Brakeman instead of an external program (e.g. time). This is roughly the time from when Brakeman reports "Processing application..." until "Generating report..."

brakeman -q --summary