I have a template function that must work with an array of float values.
something like this:

void foo(T : A[], A)(auto ref T arg)
if(is(A == float))
{
	...
}

Array may be static or dynamic. But the length of the array must be 16 or 32. How can i test length value?

On Saturday, 5 January 2013 at 21:50:19 UTC, ref2401 wrote:
> I have a template function that must work with an array of float values.
> something like this:
>
> void foo(T : A[], A)(auto ref T arg)
> if(is(A == float))
> {
> 	...
> }
>
> Array may be static or dynamic. But the length of the array must be 16 or 32. How can i test length value?

If the array can be dynamic, then the only possible way to check the array's length is with a run-time check:

if (arg.length != 16 && arg.length != 32) assert(false, "array must be 16 or 32");

That said, you can optimize both paths:

static if (isStaticArray!T)
    static assert (arg.length != 16 && arg.length != 32, "errror"); //optimized compile time check
else
    assert (/+same as above+/); //Run-time check

Depending on what you want, you can also integrate the compile time check into a conditional.

----
Note that with "T : A[]", you will accept types that aren't actually arrays, but castable to. Why not use:
"isArray!T && is(ElementType!T == float)"

Also, note your template will not work on immutable(float)[]. May or may not be what you want. Just saying.

thanks)

On Sat, Jan 5, 2013 at 10:50 PM, ref2401 <refactor24@gmail.com> wrote:

> I have a template function that must work with an array of float values. something like this:
>
> void foo(T : A[], A)(auto ref T arg)
> if(is(A == float))
> {
>         ...
> }
>
> Array may be static or dynamic. But the length of the array must be 16 or 32. How can i test length value?
>

For a dynamic array, its type is T[] and the length is a known only at runtime (by definition). You can test it with a runtime test, but not with a template constraint.

For a static array T[n], n is part of the type so you can indeed filter the
bad ones during compilation.
Using std.traits and std.range:

import std.array;
import std.traits;
import std.range;

void foo(T)(auto ref T arg)
if (  isDynamicArray!T && is(ElementType!T == float)
   || isStaticArray!T && (T.length == 16 || T.length == 32) &&
is(ElementType!T == float))
{}

void main()
{
    float[] dynamicArr;
    float[3] staticArr1 = [0.0,1.0,2.0];
    float[16] staticArr2;

    foo(dynamicArr);
    //foo(staticArr1); // fails
    foo(staticArr2);
}

Or, if you prefer using the is() expression:

void bar(T)(auto ref T arg)
if (  is(T _ == U[], U)
   || is(T _ == U[n], U, size_t n) && (n == 16 || n == 32))
{}

void main()
{
    float[] dynamicArr;
    float[3] staticArr1 = [0.0,1.0,2.0];
    float[16] staticArr2;

    bar(dynamicArr);
    //bar(staticArr1); // fails
    bar(staticArr2);
}