@picofarad if you want to be really silly, make a python script that just calls ripgrep and one that just calls grep. then have hyperfine benchmark those. it should eliminate the interpreter overhead from the comparison (by imposing it on everything.)