Ask HN: Quantitative tests/analytics frameworks?
Particularly as someone involved with research, I'm interested in a framework to manage tests which produce a quantitative value or graphs measuring performance, and compare them across branches/old commits (I'm hoping it will work with Git.) My main use case is to tweak an algorithm and keep variants in multiple Git branches, and generate some graphs measuring performance on each branch. Testing old commits should be helpful for regression testing too, especially if one test runs some unit tests and returns the runtime/memory usage (to track performance regressions). I'm hoping it will also be usable for less technical scientific programmers, to encourage good software practices.
Ideally this would work as a Git post-receive hook on a build server, maybe including an HTTP server with some interactive visualizations. Does anything similar exist, or is there something useful to build off of? I don't want to reinvent the wheel, but this is something I'm interested in building otherwise.
No comments yet.