Potential bug in __traits(parent).

[Mike]

On Saturday, 29 March 2014 at 03:00:30 UTC, Mike wrote:
> Consider the following code:
>
> /*********** mmio.d ************/
> module mmio;
>
> mixin template HasValue(uint value) {
>     static uint getValue() {
>         return value;
>     }
> }
>
> mixin template Parent() {
>     final abstract class Child(uint value) {
>         mixin HasValue!(value);
>     }
> }
>
> /*********** test.d ************/
> module test;
> import mmio;
>
> mixin Parent;
>
> alias Two = Child!(2);
>
> /*********** main.d ************/
> import std.stdio;
> import test;
>
> void main()
> {
>     writeln(__traits(parent, test.Two).stringof);
>     writeln(__traits(parent, __traits(parent, 
> test.Two)).stringof);
>
>     readln();
> }
>
> DMD 2.065 Output:
> Child!(2)
> Child!(2)
>
> GDC 4.8.2 20130725 Output:
> Child
> Child
>
> LDC 0.12.1 Output:
> Child
> Chlld
>
> All of these seem wrong to me.  What I expect for output is:
> module test
> {the second `writeln` shouldn't compile since `test` has no 
> parent}
>
> A couple of observations:
> 1. `Child` is not the parent of `test.Two`, `test` is the 
> parent of `test.Two`.
> 2. If you believe `__traits(parent, test.Two)` is `Child`, 
> `Child` should not be its own parent.
>
> Incidentally, if I remove the template parameter on `Child`, I 
> get the results I expect.
>
> Could someone please confirm if this is a bug or not?
>
> Thanks for the help,
> Mike

https://d.puremagic.com/issues/show_bug.cgi?id=12496