Re: Zip v8, stuff on compression?
Thu, 23 Nov 1995 17:50:17 -0500 (EST)
In article <GDR11.95Nov17185932@stint.cl.cam.ac.uk>,
Gareth Rees <gdr11@cl.cam.ac.uk> wrote:
>Dark Fiber <entropy@I_should_put_my_domain_in_etc_NNTP_INEWS_DOMAIN> wrote:
>> The compression in ZIP is shithouse. its practically non-existant, and
>> i feel it needs to be replaced with something better. Obviously such
>> techniques as LZW and HUFFMAN is not worthwhile as each string is
>> individually compressed, and RLE encoding will gain us nothing.
>
>The compression in ZIP, with good choice of abreviations, is about
>60-65 Your compression will do 50at best, less if there are odd
>pairs of letters, so maybe 55 I don't think that the 5-10 0ifference
>is worth the bother that would result from changing the design of the
>Z-machine.
>
>In any case, if you're worried about disk space, then use gzip.
>
>--
>Gareth Rees
Let's try this again...
Seeing an excellent opportunity to avoid working on more important
projects, I wrote a short C implementation of Fiber's algorithm (which
is included below). Using some messages from an old game source file
of mine, I found that his algorithm compresses messages down to about
60-700f their original size, which is roughly comparable to Radix-40,
the compress scheme used by Inform and the various Zip interpreters.