DMD producing huge binaries

[krzaq via Digitalmars-d]

On Saturday, 21 May 2016 at 17:34:19 UTC, Kagamin wrote:
> On Friday, 20 May 2016 at 19:37:23 UTC, Andrei Alexandrescu 
> wrote:
>> Was talking to Walter on the phone and he just had one of 
>> those great ideas: encode in the function mangle that it 
>> returns "auto". I thought that was fantastic. Would that work? 
>> -- Andrei
>
> Equivalent to not mangling return type at all. But this leaves 
> you with 2^^n growth, still exponential. Also is there any 
> evidence that compression is faster than hashing?

Would it be practical to use type-erasure in the middle of a call 
chain? It would cut 2^^n right at the knees at the possibly 
negligible runtime cost.

It would impose a burden on the client (as I don't think a 
compiler should do it on its own - or even if it could), but if 
it's really Pareto distributed then it shouldn't be that 
difficult to find the hot spots.