DMD producing huge binaries

[Walter Bright via Digitalmars-d]

On 5/21/2016 1:09 PM, Era Scarecrow wrote:
>  Depends on implementation and algorithm. However even the weakest compression
> settings can yield huge initial compression benefits. In normal text a reduction
> of 2:1 is expected, and will . The benefit being compression still contains all
> it's information, and hashing will reduce the size to a fixed length but throw
> away all that information for unique identity.
>
>  LZO is a compressor/decompressor that prioritizes speed and can be used in
> real-time applications for very little cost to add compression to any
> application. I tinkered with some of it's options a while back and it did okay,
> not as good as zlib but that's to be expected. Although using zlib on it's
> fastest setting will likely yield good results too.
>
>  http://www.oberhumer.com/opensource/lzo/

We already have a compressor in the compiler source for compressing names:

   https://github.com/dlang/dmd/blob/master/src/backend/compress.c

A faster one would certainly be nice. Anyone game?