2.065 compiler problem

[monarch_dodra]

On Thursday, 3 April 2014 at 07:38:33 UTC, monarch_dodra wrote:
> On Thursday, 3 April 2014 at 07:05:13 UTC, Ondrej Pokorny wrote:
>> Same under windows. Is this compiler/phobos bug?
>
> I can tell *why* it's breaking right now, but have been unable 
> to understand what caused said breakage, or how it worked at 
> all to begin with.
>
> Will reduce and file (and potentially fix).

I fixed the implementation so that your test passes now. The 
implementation was... odd. I think it may have been "legacy code" 
from before we had "allSatisfy" and "anySatisfy" to easilly do 
introspection.

FYI, the issue was that "RepresentationTypeTuple" was eagerly 
called on input, regardless of type. Because you were operating 
on a forward reference, the RepresentationTypeTuple failed.

My fix consisted in calling RepresentationTypeTuple *after* first 
checking the current type.

I will submit the fix shortly.