PAsQDa 4.4 (03.01.2006) - File archiver and compressor (based on PAQAR 4.1).
-----------------------   (C) 2005, M. Mahoney (mmahoney@cs.fit.edu), F. Buffoni
                                A. Ratushnyak (artest@inbox.ru), B. Destasio
                                P. Skibinski (inikep@o2.pl), D. Scott, J. Schmidt
                          http://www.ii.uni.wroc.pl/~inikep/research/pasqda44.zip


1. Introduction
PAsQDa is PAQAR with added a preprocessor (WRT by P. Skibinski), which
improves the compression performance, particularly on English texts.

WRT (http://www.ii.uni.wroc.pl/~inikep/research/WRT44.zip) is a high-performance
file to file preprocessor NOT ONLY for textual files. It is based on StarNT,
improved version of LIPT. The main idea behind these algorithms is to replace
words with codewords (numbers) in the dictionary.

To replace we use ASCII characters from 128 up to 255 as they are rarely
used in text files. We use 4 subdictionaries, which outputs from 1 to 4
characters. Most frequent words are in the first subdictionary. Often used
words are in the second subdictionary and rarely used words are in the
third and the fourth subdictionary. 



2. Usage
To compress:      PAsQDa -5 archive file file...  (1 or more file names), or
  or (MSDOS):     dir/b | PAsQDa -5 archive       (read file names from input)
  or (UNIX):      ls    | PAsQDa -5 archive
To decompress:    PAsQDa archive                  (no option)
To list contents: more < archive 

The -5 is default, and is used to trade off compression vs. speed
and memory.  Valid options are -0 to -9.  Higher numbers compress better
but run slower and use more memory. 



3. History
PAsQDa 4.4 (03.01.2006)
-improved file type detection routines
-better compression of multi-lingual text files

PAsQDa 4.3c (19.12.2005)
-small bug fix in a preprocessor (for files with a punctuation mark as the last character)

PAsQDa 4.3b (9.12.2005)
-bug fix for executables compressed with the -e option (thanks to A. Ratushnyak for notification)
-small bug fix in a preprocessor (for files with a space as the last character)

PAsQDa 4.3 (07.12.2005)
-PASQDA.EXE: default settings as in PAsQDa 4.1/4.1b
-"-i" option (from -i0 to -i10, default: 0 and 8 for -Me), which changes update weight of even mixers
-"-j" option (from -j0 to -j20, default: 2 and 20 for -Me), which changes update weight of odd mixers
-"-w" option (from -w0 to -w32, default: 16 and 28 for -Me)
-compiled on Intel 9 (thanks to Johan De Bock)

PAsQDa 4.2 (05.12.2005)
-"-w" option, which changes balance between prediction and SSE (from -w0 to -w32)
-lower memory requirements (32 MB less for -6 option)
-bug fix for 16-bit executables compressed with the -e option

PAsQDa 4.1b (01.10.2005)
-bugfix in preprocessor (thanks to A. Ratushnyak for notification)

PAsQDa 4.1 (01.07.2005)
-PAsQDa split into two versions: PAsQDa (low memory consuming) and PAsQDaCC (optimized for Calgary Corpus)
-changed Capital Conversion flag and Escape flag mapping
-EOL coding: special model for CR+LF+Space and LF+Space

PAsQDa 4.0 (31.03.2005)
-changed dictionary mapping
-changed Capital Conversion flag mapping
-new dictionary
-slightly improved EOL coding

PAsQDa 3.0 (07.02.2005)
-word model optimized for preprocessor
-during compressing the Calgary Corpus:
  a) charModel trained on the file book2 becomes a predictor for textual files
  b) memory requiremens are higher (run "pasqda.exe" without parameters)

PAsQDa 2.1 (31.01.2005)
-automatic restart of PAQ model on non-textual files (in multi-file compression)
-non-textual files are compressed without WRT-based preprocessor

PAsQDa 2.0 (24.01.2005)
-compiled with Visual C++ 7.1 (.NET 2003)
-based on PAQAR 4.0
-sorts input files depending on kind of file

PAsQDa 1.0 (18.01.2005) 
-first public release
-compiled with Visual C++ 6.0
-based on PAQ6 v2



4. Licence
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License version 2 as
published by the Free Software Foundation at
http://www.gnu.org/licenses/gpl.txt or (at your option) any later version.
This program is distributed without any warranty.
