DMD producing huge binaries

[poliklosio via Digitalmars-d]

On Saturday, 21 May 2016 at 08:57:57 UTC, Johan Engelen wrote:
> On Saturday, 21 May 2016 at 06:18:05 UTC, poliklosio wrote:
>>
>> I have an Idea of reproducible, less-than-exponential time 
>> mangling, although I don't know how actionable.
>>
>> Leave mangling as is, but pretend that you are mangling 
>> something different, for example when the input is
>> foo!(boo!(bar!(baz!(int))), bar!(baz!(int)), baz!(int))
>> pretend that you are mangling
>> foo!(boo!(bar!(baz!(int))), #1, #2)
>> Where #1 and #2 are special symbols that refer to stuff that 
>> was **already in the name**, particularly:
>> #1: bar!(baz!(int))
>> #2: baz!(int)
>
> See 
> http://forum.dlang.org/post/szodxrizfmufqdkpdryc@forum.dlang.org

So if I understand correctly, you tried implementing something 
like this, but it didn't help much even with size of mangled 
names. Are you sure you were testing on the pathological case 
(exponential stuff), rather than a bad one? Assuming your 
experiment is correct, something interesting is happening, and I 
will be observing you guys finding out what it is. :)