We can pass reference of an array to a function like:
void f(int (&a)[5]);
int x[5];
f(x); //okay
int y[6];
f(y); //error - type of y is not `int (&)[5]`.
Or even better, we can write a function template:
template<size_t N>
void f(int (&a)[N]); //N is size of the array!
int x[5];
f(x); //okay - N becomes 5
int y[6];
f(y); //okay - N becomes 6
Now my question is, how to return reference of an array from a function?
I want to return array of folllowing types from a function:
int a[N];
int a[M][N];
int (*a)[N];
int (*a)[M][N];
where M
and N
is known at compile time!
What are general rules for passing and returning compile-time reference of an array to and from a function? How can we pass reference of an array of type int (*a)[M][N]
to a function?
EDIT:
Adam commented : int (*a)[N]
is not an array, it's a pointer to an array.
Yes. But one dimension is known at compile time! How can we pass this information which is known at compile time, to a function?
If you want to return a reference to an array from a function, the declaration would look like this:
// an array
int global[10];
// function returning a reference to an array
int (&f())[10] {
return global;
}
The declaration of a function returning a reference to an array looks the same as the declaration of a variable that is a reference to an array - only that the function name is followed by ()
, which may contain parameter declarations:
int (&variable)[1][2];
int (&functionA())[1][2];
int (&functionB(int param))[1][2];
Such declarations can be made much clearer by using a typedef:
typedef int array_t[10];
array_t& f() {
return global;
}
If you want it to get really confusing, you can declare a function that takes a reference to an array and also returns such a reference:
template<int N, int M>
int (&f(int (¶m)[M][N]))[M][N] {
return param;
}
Pointers to arrays work the same, only that they use *
instead of &
.