wild question re compression

Archiver

Here is a wild question. I need a compression algorythm that is agressive on small ammounts of data (44-50) of the charicters A-Z a-z and 0-9. Huffman is "ok" but does anyone know of a better tequniuqe? Anything with the stamp's limitation is prefered but even C code is helpful. Thank you for the help.
MH

Archiver

I did an article a long time ago for Visual Developer on Base 40
compression. The idea is to make your symbols fit in 40 characters (usually
A-Z 0-9 space and 3 punctuation of your choice). Now you can fit 3
characters in two bytes using:

Word = C2*40**2 + C1*40**1 + C0*40**0
** is exponentiation which you precompute to be:
C2*1600+C1*40+C0

Then you can split them apart just like you split a hex number (but using
40) / is integer division:

C2=Word/1600
C1=(Word-C2*1600)/40
C0=Word//40 ' remainder operation

Why not 41? Because

39*1600+39*40+39=63999 (<2**16)
but
40*1681+40*41+40 = 68920 (>2**16)

This is an old mainframe trick which goes to prove that if you remember
things long enough you can make them new again :-)

Earthweb bought the rights to all the old Visual Developer articles when
Coriolis shut it down, so you might find the article on their Web site (it
was called Pack it Up I think) Try www.developer.com
They have a lot of my old columns online (my wife likes to read the stories
and then quits when they get technical). But I didn't see that one online
off the top of my head.

Regards,

Al Williams
AWC
* Floating point math for the Stamp, PIC, SX, or any microcontroller at
http://www.al-williams.com/awce/pak1.htm

---

Original Message

---

From: Michael Hendricks [noparse]/noparse]mailto:[url=http://forums.parallaxinc.com/group/basicstamps/post?postID=KLrteXoMnQ1y0j2OkEW9rk51NgKNFopqSyaG5hsTVO99w9815quCzUHZMO4qI17zK1Mwfq3UmOYPlull]mjh80@b...[/url
Sent: Monday, November 06, 2000 10:54 PM
To: basicstamps@egroups.com
Subject: [noparse][[/noparse]basicstamps] wild question re compression


Here is a wild question. I need a compression algorythm that is agressive on
small ammounts of data (44-50) of the charicters A-Z a-z and 0-9. Huffman is
"ok" but does anyone know of a better tequniuqe? Anything with the stamp's
limitation is prefered but even C code is helpful. Thank you for the help.
MH