So if the local variable can be changed, then we can use loop, etc. same as imperative languages. For example, for (i=0; i<100; i++) where `i` is a local variable in function.
It is true that a pure language could support such things (some pure languages do, e.g. Sisal). The experience of people using SML and Haskell seems to indicate that this is not very important in those languages: it's easy for the programmer to turn his variable-assignments into something pure (using new variables instead), and it's reasonably easy as well for the compiler to recognize loops and handle them just as efficiently as if they were while/for loops in an imperative language. OTOH, allowing modification of variables makes the whole language impure, because of variables caught in closures, e.g.: let count = 0 in \() -> count := count + 1; count so it has very far reaching consequences. Stefan