Hi, I've been trying to reduce a bug in the containers (8824).

From the example below it seems the dup method is passing the constructor
an array of dchars and the template is failing.

Is this a compiler bug, or ?

import std.range, std.traits;

struct Array2(T)
{
    private T[] _payload;

    size_t insertAt(R)(size_t index, R rng)
        if (isInputRange!R && isImplicitlyConvertible!(ElementType!R, T))
    {
        return 0;
    }

    this(U)(U[] values...)
        if (isImplicitlyConvertible!(U, T))
    {
        insertAt(0, values);
    }

    @property Array2 dup()
    {
        return Array2(_payload);
    }
}

unittest
{
    Array2!int a; //passes
    Array2!char a; //fails
}

On Sunday, 24 August 2014 at 02:53:41 UTC, Damian Day wrote:
> Hi, I've been trying to reduce a bug in the containers (8824).
>
> From the example below it seems the dup method is passing the constructor
> an array of dchars and the template is failing.
>
> Is this a compiler bug, or ?
>
> import std.range, std.traits;
>
> struct Array2(T)
> {
>     private T[] _payload;
>
>     size_t insertAt(R)(size_t index, R rng)
>         if (isInputRange!R && isImplicitlyConvertible!(ElementType!R, T))
>     {
>         return 0;
>     }
>
>     this(U)(U[] values...)
>         if (isImplicitlyConvertible!(U, T))
>     {
>         insertAt(0, values);
>     }
>
>     @property Array2 dup()
>     {
>         return Array2(_payload);
>     }
> }
>
> unittest
> {
>     Array2!int a; //passes
>     Array2!char a; //fails
> }

Not a compiler bug. char[] (and wchar[]) as ranges auto-decode to
dchars. I.e. ElementType!(char[]) is dchar. And dchar isn't
implicitly convertible to char. So the instantiation of insertAt
fails.

On Sunday, 24 August 2014 at 02:53:41 UTC, Damian Day wrote:
> isImplicitlyConvertible!(ElementType!R, T))

Try [ElementEncodingType][0].

[0]: http://dlang.org/phobos/std_range.html#ElementEncodingType