In one of the linked posts he says that 100% test coverage might not be truly achievable, but it's nonetheless a valid asymptote to aspire to. I think that exa

In one of the linked posts he says that 100% test coverage might not be truly achievable, but it's nonetheless a valid asymptote to aspire to. I think that exact argument applies to designing ever-safer, ever-better languages. You might never have the perfect tools that ensure 100% program correctness, but it's sure as hell what a tool-maker should aspire to.