這里是LDS(lossless data-compression sources)的KIT.我發現很多人都想這么一份壓縮軟件的源代碼,但是他們很難找到,特別是一些比較好的.
    所有的源代碼已經用Borland C++編譯成功.每個目錄中都包含一個makefile或者是make.bat用以創建可執行的文件.其它的編譯器或操作系統需要改動一些地方.
    所有的源代碼包括它們的原始文檔.





=========

AR002
   Contains a complete archiver by Haruhiko Okumura. The archiver uses
   an LZ engine whose output is compressed with static Huffman. The
   file is chopped into pieces which all get their own independent
   set of Huffman tree's. ARJ 2.41, ZIP 2.0 and PKZIP 2.04g use
   essentially the same method. ZOO and LHA even use excactly the
   same method.

LZW
   The famous Lempel Ziv Welch compressor. This version by Kent Williams.

COMP
   Arithmetic compression written by Mark R. Nelson. Published in
   Dr Dobbs february 1991. Contains a basic arithmetic compressor
   and a higher order arithmetic compressor.

ASH
   A "brother" of the higher order compressor from Nelson written by
   Charles Ashford. Ingenious program which makes use of the very limited
   memory of the PC and still can process large files. One price to pay,
   it is VERY slow. The tigthest compressor in the Dr Dobbs compression
   contest.

FIN
   Finish submission to the Dr Dobbs contest written by Jussi Puttonen,
   Timo Raita and Jukka Teuhola. The fastest DOS compressor I know of.
   The origional is in 100% assembly. I derived a C version (which I
   published before in PC Newsflash).

HUF
   Static Huffman compression/decompression by William Demas.

LZ
   LZSS by Haruhiko Okumura. LZARI by Haruhiko Okumura, the first
   compressor recompressing the output of the LZ engine. Haruyasu
   Yoshizaki (the author of LHarc and LHA) replaced the ARI compressor
   by dynamic Huffman, LZHUF.

SPLAY
   Splay tree compression written by Kim Kokkonen. Splay tree compression
   is a "nephew" of Huffman compression. It is faster but achieves less
   compression in most cases.

LZRW1
   LZRW1 compressor by Ross Williams. A very fast LZ compressor. Not as
   fast as FIN but it has a better compression ratio. I appended a small
   part to show it really works. This implementation of LZRW1 uses memory
   as source and destination, not disk.