From: Daniel Martin <martin@...>
Date: 2006-08-09T22:39:13+09:00
Subject: Re: Codegolf - Writing a Brainf*ck interpreter

Michael Ulm <michael.ulm@isis-papyrus.com> writes:

> Very nice. If you accept a more fragile version (but still conforming
> to the specs) you can bring that down to 210 bytes:
>
> c,$k=STDIN.read.split'!'
> j=%w[d="\0"*8**5 a=0]
> c.each_byte{|i|j<<({10,"",?<,"a-=1",?>,"a+=1",
> ?[,"while d[a]>0 do",?],"end",?.,"print d[a,1]",
> ?,,"d[a]=$k.slice!(0)||exit"}[i]||"d[a]+=#{44-i}")}
> eval j.join("
> ")

I managed to squeeze out a few more characters by replacing the hash
lookup with an array index.  I also switched out slice!(0) for an
indexing operation, but in the end that caused no overall character
count change.  Anyway, here's 198 characters:

c,$k=STDIN.read.split'!'
j=%w[d="\0"*8**5 q=a=0]
c.each_byte{|i|j<<'
a-=1
a+=1
while d[a]>0
end
print d[a,1]
d[a]=$k[q]||exit;q+=1
d[a]+=1
d[a]-=1'.split('
')["
<>[].,+-".index(i)]}
eval j.join("
")

As an added bonus (?), all the newlines in that are necessary to the
code.

Still, there's supposedly another way to squeeze an additional 50
characters out, which I think is going to require some sort of
fundamental shift in the approach to the problem.