Post #2466484
2026-05-04 21:11 UTC
@glloyd@fosstodon.org @dascandy@infosec.exchange Right. I was so enthusiastic about mutation testing after watching Seph Debusser's talk from NDC TechTown.
https://www.youtube.com/watch?v=M-5_M8qZXaE
But then I kind of forgot, I guess.
I imagine it's great when writing new tests to ensure that you really do cover what you intend, but I fear that trying it on (in this case 12 year) old code, may be more of a time sink than value?
Experiences?
Replies (1)
-
@dascandy@infosec.exchange 2026-05-04 21:30
@rollbear@fosstodon.org @glloyd@fosstodon.org It likely does more useful than the attempt I once made to measure MCC coverage within the unit tests themselves, to find a set of unit tests that would get the same coverage but with less tests to run. I got it to work. It's a terrible idea. It'll tell you that your fourier transform tests can be replaced with a single test of zero bytes being put in, since that gives the same coverage.