I had no idea that people were quite so religious about this sort of thing. It’s pretty clear at this point that testing is one of the most valuable tools in t

I had no idea that people were quite so religious about this sort of thing. It’s pretty clear at this point that testing is one of the most valuable tools in the box for getting sufficiently “correct” software in most domains. But it’s only one tool. Some people would call property checkers like Hypothesis or QuickCheck “testing”, some people wouldn’t. Either way they are awesome. Formal methods are also known to be critical in extreme low-defect settings, and seem to be gaining ground more generally, which is a good thing. Richer and richer type systems are going mainstream with like Rust and other things heavily influenced by Haskell and Idris et al. And then there’s good old: “shipping is a feature, and sometimes a more important feature than a low defect count”. This is also true in some settings. jwz talk very compellingly about this. I think it’s fine to be religious about certain kinds of correctness-preserving, defect-preventing processes in domains that call for an extreme posture on defects. Maybe you work on avionics software or something. But in general? This “Minimal test case! Red light! Green light! Cast out the unbelievers!” is woo-woo stuff. I had no idea people took this shit seriously.