11 Mar
2025
11 Mar
'25
12:23 a.m.
Gnu lzip -9 got a 3.92 compression factor , while Morris-Thompson got 4.52.
According to the lzip man page, lzip has integrity checks, which must
take some space. It is also said to be a "simplified" L-Z, though the
only simplification that is mentioned--bounding the size of the
dictionary--can only happen on much bigger files than Webster. If you
still wonder how well "pure" L-Z does, I'll be happy to try your
favorite implementation.
It did not surprise me that lzip lost to M-T, for it has to discover
over and over again that the dictionary is lexicographically
ordered--a fact that gives M-T most of its leverage. The dictionary is
definitely not generated by a Markov process!
Doug
On Mon, Mar 10, 2025 at 6:56 PM James Johnston <audioskeptic(a)gmail.com> wrote:
Doug, do you know how much better Ziv-Lempel does? I'm guessing "a little
bit"?
On Mon, Mar 10, 2025 at 3:52 PM Douglas McIlroy <douglas.mcilroy(a)dartmouth.edu>
wrote:
I have sent a copy to Warren for the Unix archives.
Doug
On Sat, Dec 28, 2024 at 9:54 PM Royce Williams <royce(a)techsolvency.com> wrote:
>
> Someone I know is seeking the original version of an internal Bell Labs memo from
1974 titled "Webster's Second on the Head of a Pin" by Morris and Thompson.
The topic appears to be related to improving the speed of lookups or search. It's
cited in a few papers as "Unpublished Technical Memo, Bell Laboratories, Murray Hill,
NJ 1974." All I can find online is citations. Any leads appreciated!
>
> --
> Royce
--
James D. (jj) Johnston
Former Chief Scientist, Immersion Networks