4 ms·
Well it's possible, but how many developer would rather make a new type for using the first ten integer instead of just using int? I think random testing is a
by StrykerKKD 10y ago
Well it's possible, but how many developer would rather make a new type for using the first ten integer instead of just using int?
I think random testing is a good way to get almost exhaustively tests.
- khedoros 10y agoA developer that uses a narrower integer data type, or who wants to be clear that the argument is a member of an enumerated set of options, not an integer. What kind of developer represents non-integer data as an integer data type? > I think random testing is a good way to get almost exhaustively tests. I think that testing known/predictable edge and corner cases makes more sense (in some cases), but I also think that I was responding to a comment about the impossibility of using tests to prove correctness of a function, not about how to derive practical benefits despite theoretical limitations.
- StrykerKKD 10y agoYou are right, I didn't think the example through. You are right, you can use tests to prove a function if - the function has very limited arguments and - the function is pure or does very limited amount of state change(side effect) Generally speaking tests are not the right tool for proving program correctness.
- 8note 10y agointegers are frequently used to represent fixed point numbers anytime you're doing interesting math in an embedded system
- khedoros 10y agoAnd old PC hardware, where you aren't guaranteed to have an FPU, or where integer operations were a lot quicker than FP ones.