🚀 go-pugleaf

In article <ebladeD4...@netcom.com>, Eric Blade <ebl...@netcom.com> wrote:
>In article <3j78t4$4...@eldborg.rhi.hi.is>,
>Hordur Kvaran <hor...@rhi.hi.is> wrote:
>>In <3inlj1$2...@crl12.crl.com> ta...@crl.com (Joseph C. Hiller) writes:
>>
>>>Adam J Lewandowski (alew...@uceng.uc.edu) wrote:
>>
>>I thought Quake wouldn't have any hard-coding.  If they are going to hard-code
>>it in some assembler how are they going to port it to different machines?  
>>
>
>  C compilers have traditionally supported an option that dumps the
>assembly code it generates to a file. I would imagine that they would get
>a working version of the client, compile it to assembly, optimize
>the living piss out of it (GO MICHAEL!) and then assemble it.
>
  Actually, if Abrash is doing the optimizing, they most emphatically
_won't_ be doing this.  Check out the section in his book, the _Zen
of Optimization_, "Billy, don't be a compiler."  He emphasizes
abstraction, and methodology level optimization over straight code
munching.

							...Dave

--
                  Dave Moore - Student at Large - Reed College
                 dmo...@reed.edu - http://www.reed.edu/~dmoore/
              Mail me with subject "request pgp" for PGP 2.6.2 key