Re: [Hampshire] Fw: Re: [hants] compression

Top Page

Reply to this message
Author: Jim Kissel
Date:  
To: isaacclose, Hampshire LUG Discussion List
Subject: Re: [Hampshire] Fw: Re: [hants] compression


Isaac Close wrote:
>> <paul@???> wrote:
>>
> > What is the nature of the material you are trying to
> > compress and what is the reason for wanting to
>> compress it
> > please?
> >
>
> Thats a good, but obvious question, unfortunately the
> answer is somewhat naive.
>
> Arbitrary (but small (maybe upto 1MB)) stuff, and that
> could mean already compressed data, we would like to know
> simply if its possible to better 7zip or bzip really.
>
> I have my doubts.
>

When tailored for the data set Huffman encoding can and does produce
better results that either 7zip or bzip. Unfortunately, to get this
compression on a set of diverse data sets requires many custom encoding
schema and a custom "program" for each data set. The overheads may be
too great for the benefits gained by the compression.

--
Life is too short.