jemarch's suggestions for configurability in the diff / regression analysis - status of outcome codes (e.g. is 'UNSUPPORTED' considered a regression?) - subset of expfiles (e.g. we care about essential-thing.exp but not weird-flaky-test.exp) Most analysis settings could be changed by tweaking the scripting. However, as jemarch points out, this requires either modifying the sourceware bunsen instance or cloning the bunsendb git repo and running your own analysis. It's beneficial to have an easy procedure to let the sourceware bunsen-analysis be tweaked for common settings per-project. Any other things deserving of easily accessible knobs?