|
|
|
|
|
by jerf
5637 days ago
|
|
All code is a trivial recombination of working parts. It's all CPU opcodes in the end. Haskell's guarantees may extend a bit further but if your program isn't trivial it still needs testing, if only to verify that the program does what you think it does. I'm fiddling around with porting my HTML normalizer into Haskell, and even if the type system reliably lets me say "Hey, this has no type errors" and gives me some reasonably assurance based on the way it is written that it has no infinite loops (not a proof, but there's not very many places they could hide), I still need tests just to make sure it does what I think it does. A lot of them, actually. |
|