Projects / BitComp / Comments

Comments for BitComp

18 Aug 2000 07:48 idanso

Just to make things clear

I got a few responses regarding similarity between the description of the algorithm to Huffman algorithm. I doubt If many of those critics took the time to look at the PHP source however

In order to make a few things clear, I don't think there is any similarity to Huffman, maybe at the very basic concept level, but not in the algorithm itself(Huffman uses B-Trees, And can compress files that use even 99% of the ASCII table. the last time i checked, BitComp did not use B-Trees, and requires files to use less than 50% of the ascii table for minimal compression), and not in the compression ratio

As a proof, i downloaded Huffman implementation from the internet(which came with the source). I compressed a sample file using both Huffman and Bitcomp. results are as follows

Original : 15096 bytes
BitComp : 13209 bytes
Huffman : 8973 bytes

So, as you can see, BitComp is not huffman. the sample file used 86/256 from the ascii table, that means BitComp encoded it in 7-bits. Huffman definatly used some other method

This does not mean BitComp is a poor compression method however, As not all features(like LZ encapsulation) are currently implemented, so while(as i claimed on the web site) it is currently easily bypassed by LZ7? algorithms, it doesn't mean it will not be improved in the near future

Hope that made a few things clear

idan

17 Aug 2000 18:12 egnor

Hmm.
Someone needs to explain Huffman coding to this person, and point out that any respectable file compression utility ('compress', gzip, bzip2, etc.) will certainly achieve the basic symbol-space compression "BitComp" touts as a "new compression technique".

Screenshot

Project Spotlight

ReciJournal

An open, cross-platform journaling program.

Screenshot

Project Spotlight

Veusz

A scientific plotting package.