4 ms·

Statically typed languages impose unnatural hardware-oriented constraints on your business logic - It forces you to spend extra time and effort to make sure tha

Statically typed languages impose unnatural hardware-oriented constraints on your business logic - It forces you to spend extra time and effort to make sure that your logic is in a format which the compiler can understand. Yes, this can sometimes help you to find silly errors in your code at compile time (and thus occasionally save you a bit of time) but, unfortunately, that occasional benefit doesn't make up for the productivity loss incurred from constantly trying to 'please the compiler'. Dynamically typed languages offer fluidity in your logic where it is needed. For example, if you add an integer and a decimal (floating point) number together, there is nothing wrong about this mathematically, but a static compiler will complain! It's a limitation of the compiler not the programmer! Statically typed languages force you to implement rigid class/type schemas within your logic (sometimes with complex hierarchies) - These schemas have to be significantly reworked whenever your business requirements change.

muraiki 12y ago
I'm not trying to be insulting, but did you read the article? From what you've written, it sounds like you haven't actually used an expressive type system. For instance, in Haskell I can type 2 + 3.75 and I get 5.75 right back -- no errors. That's because the type signature of + is "Num a => a -> a -> a," meaning I can add two of any type of Number (Integers and Floats are both Numbers, of course).
ufo 12y ago
Actually, I see having to rework the type schema when the requirements change as a plus, not as a con. In a static language the compiler can point out all the bits of the program that need to be updated to conform to the new schema while in a dynamic language this kind of stuff is usually only detected at runtime.
theseoafs 12y ago
Have you never used a statically-typed language? Most statically typed languages can handle adding an integer and a float together. The only one I can think of that isn't capable of that is OCaml, though this is probably a strength of OCaml rather than a weakness (int-to-float casting can be a common source of bugs if it happens unexpectedly).
- ufo 12y ago
  In Ocaml's case its less about casting being a source of bugs and more about casting not playing well with Hindley-Milner type inference, which is Ocaml's "killer feature". (and Ocaml doesn't have type classes to solve this problem like Haskell does)
- jondubois 12y ago
  Try to add an int and a float32 using 'Try go' on http://glolang.org http://glolang.org - I get this error: prog.go:13: invalid operation: x + y (mismatched types int and float32) [process exited with non-zero status] I'd have to cast one of the variables to get it to work. Casting is unnatural - Maybe it's not so bad in this case, but in many cases it isn't desirable.
codygman 12y ago
Your comment certainly doesn't apply to languages with expressive type systems.
Peaker 12y ago
I think you should spend some time familiarizing yourself with actual statically typed languages before expressing an extremely ill-informed opinion as fact.

muraiki 12y ago

I'm not trying to be insulting, but did you read the article? From what you've written, it sounds like you haven't actually used an expressive type system. For instance, in Haskell I can type 2 + 3.75 and I get 5.75 right back -- no errors. That's because the type signature of + is "Num a => a -> a -> a," meaning I can add two of any type of Number (Integers and Floats are both Numbers, of course).

ufo 12y ago

Actually, I see having to rework the type schema when the requirements change as a plus, not as a con. In a static language the compiler can point out all the bits of the program that need to be updated to conform to the new schema while in a dynamic language this kind of stuff is usually only detected at runtime.

theseoafs 12y ago

Have you never used a statically-typed language? Most statically typed languages can handle adding an integer and a float together. The only one I can think of that isn't capable of that is OCaml, though this is probably a strength of OCaml rather than a weakness (int-to-float casting can be a common source of bugs if it happens unexpectedly).

ufo 12y ago
In Ocaml's case its less about casting being a source of bugs and more about casting not playing well with Hindley-Milner type inference, which is Ocaml's "killer feature". (and Ocaml doesn't have type classes to solve this problem like Haskell does)
jondubois 12y ago
Try to add an int and a float32 using 'Try go' on http://glolang.org http://glolang.org - I get this error: prog.go:13: invalid operation: x + y (mismatched types int and float32) [process exited with non-zero status] I'd have to cast one of the variables to get it to work. Casting is unnatural - Maybe it's not so bad in this case, but in many cases it isn't desirable.

codygman 12y ago

Your comment certainly doesn't apply to languages with expressive type systems.

Peaker 12y ago

I think you should spend some time familiarizing yourself with actual statically typed languages before expressing an extremely ill-informed opinion as fact.