Ah, Pascal differentiates between procedures and functions, but not in the procedural vs functional way. Pascal functions may have side effects, there is nothing in the language forbidding that.
Regarding the SQL extensions, well, the distinction between functions and procedures is not so clear. The main difference seems to be as in the Pascal case: if there is a return value or not. But then, function calls are allowed embedded inside SQL statements and that opens the door to lots of other considerations, more related to database implementation and internal matters than to anything else.
I think it's actually a little unlucky that C books speak about "void functions" rather than "procedures".
Just returning a value doesn't make a function a function in the functional programming sense (deterministic, no side effects). In C (at least) that is exacerbated by the fact that return values are commonly used to report errors.
So, it's true that C functions returning void are procedures, but then, most C functions returning values are also procedures. The distinction doesn't lie there. | [reply] |
| [reply] |
A function in the mathematical sense is something a little different than a function in the programming sense.
No, it depends on the context:
Under the imperative programming paradigm, "function" can refer to anything: subroutines with and without return values; deterministic and non-deterministic; and with and without side effects.
Under the functional programming paradigm, "function" nowadays usually has the mathematical meaning... well, at least in hardcore functional-programming circles (i.e. between Haskell programmers).
The differentiation between function and procedure a-la Pascal is, frankly, uninteresting and mostly useless.
| [reply] |