Blind Copy
I just finished watching Simon Brown’s brilliant talk: “Software Architecture vs Code”.
I thought the segment he presented on levels of testing was really, really good. Simon had the nerve to question the dogma of TDD and the dubious value of unit testing the hell out of your code (90%, 100% coverage anyone?). He cited the controversial writings of Jim Coplien and David Heinemeier-Hansson that poke some holes in those revered, extreme practices:
Like Cope and DHH, Simon does not advise shit-canning ALL unit testing. He simply suggests rethinking the test pyramid and how one allocates resources to the various levels of testing:
Instead of mandating 90 or 100 percent unit test coverage in order to create a high level of (false) confidence in your code base, perhaps you and your org should consider the potential silliness of the current obsession with TDD and writing huge unit test suites. Maybe you’d save some money and deliver your product faster. But then again, maybe not.