decoding foreach

[Timon Gehr via Digitalmars-d]

On 02.06.2016 12:38, deadalnix wrote:
>>
>
> This, deep down, point at the fact that conversion from/to char types
> are ill defined.
>
> One should be able to convert from char to byte/ubyte but not the other
> way around.
> One should be able to convert from byte to short but not from char to
> wchar.
>
> Once you disable the naive conversions, then the autodecoding in foreach
> isn't inconsistent anymore.

The current situation is bad:

void main(){
     import std.utf,std.stdio;
     foreach(dchar d;"∑")
         writeln(d); // "∑"
     foreach(dchar d;"∑".byCodeUnit)
         writeln(d); // "â", "ˆ\210", "\221‘"
}

Implicit conversion should not happen, and I'd prefer both of them to 
behave the same. (I.e. make both a compile-time error or decode for both).