The learn forum is great not only for asking questions, but also for learning from the questions by others: I just learned about the existence of std.range.primitives: ElementType, a function that I have looked for in phobos before, without finding it. I had expected this to be in std.traits or __traits.

So I implemented my own:
```
template arrayElementType(T : T[])
{
    alias arrayElementType = T;
}

unittest
{
    static assert (is(arrayElementType!(int[]) == int));
}
```

For reference, this is the phobos version:
```
template ElementType(R)
{
    static if (is(typeof(R.init.front.init) T))
        alias ElementType = T;
    else
        alias ElementType = void;
}
```

As far as I can see these are largely equivalent, except that the phobos version has special behavior for narrow strings [1]. This special behavior is actually unwanted in my case.

The following questions pop up:
1) Should we have a reference in the docs for std.traits to std.range.primitive : ElementType?
2) Should phobos contain a version without the special narrow string behavior?
3) Are there other differences between my version and the phobos version?

Thanks,
Bastiaan.

[1] https://dlang.org/phobos/std_range_primitives.html#ElementType

On Sunday, 7 January 2018 at 12:40:22 UTC, Bastiaan Veelo wrote:
> 1) Should we have a reference in the docs for std.traits to std.range.primitive : ElementType?

wouldn't hurt i guess

> 2) Should phobos contain a version without the special narrow string behavior?

see ElementEncodingType
http://dpldocs.info/experimental-docs/std.range.primitives.ElementEncodingType.html

> 3) Are there other differences between my version and the phobos version?

The Phobos one works on any range, not just built-in arrays. It also returns an answer (void) if you pass it something that has no element type, whereas yours would be a compile error.

On Sunday, 7 January 2018 at 13:50:23 UTC, Adam D. Ruppe wrote:
> On Sunday, 7 January 2018 at 12:40:22 UTC, Bastiaan Veelo wrote:
>> 1) Should we have a reference in the docs for std.traits to std.range.primitive : ElementType?
>
> wouldn't hurt i guess
>
>> 2) Should phobos contain a version without the special narrow string behavior?
>
> see ElementEncodingType
> http://dpldocs.info/experimental-docs/std.range.primitives.ElementEncodingType.html
>
>> 3) Are there other differences between my version and the phobos version?
>
> The Phobos one works on any range, not just built-in arrays. It also returns an answer (void) if you pass it something that has no element type, whereas yours would be a compile error.

Thanks.