huffman
                                  -------
          Documentation for version 1.0, on September 11th, 1991

                      by Bruno Costa and Lucia Darsa



WARNING

  Although  this  program  is freely distributable (see COPYRIGHT below), I
would  appreciate  if  you  take the time to write me a line or two (E-Mail
please)  if  you  are  using  any of my programs or if you have any kind of
ideas  or  suggestions  (This is not too much to ask, is it?).  If you feel
particularly  generous,  I  would  like  to  receive a disk with any useful
programs you have written, preferrably with source.  Thanks!


DESCRIPTION

  Huffman  is  a  (not so) simple program to compress files, using the well
known  huffman  coding  algorithm.  The compression ratios achieved by this
method  are worse than those obtained by compress, arc, lharc, zoo, zip and
many  others,  but at least you have the source code ;-).  Huffman has been
ported  easily  to  MeSsy-DOS  and UNIX, and can be used as a simple way to
compress  files to be transferred between different systems, over which you
have  control  (to  remove  bugs or add features).  Huffman works perfectly
under  AmigaDOS  1.3,  but  under  2.0  you  must disable fast memory using
NoFastMem  (don't  ask  me why!), as is necessary to run PKAZIP.  Note that
this  was  tested in an Amiga 3000 under Kickstart v36.209, with burst mode
and caches turned either on or off.


OPTIONS

  usage: huffman [-c | -u | -r] [-f | <files>...]

  You  must select one and only one of the 'c', 'u' or 'r' options.  Option
'c'  makes  huffman compress the given files, and 'u' is used to uncompress
them.  To just estimate the compression ratio, use 'r'.

  Option 'f' is used to make huffman work as a filter (not implemented).

  You may give as many files as you wish, huffman will compress each one in
turn, and if the compression is successful, it will replace the old file
with the compressed one, keeping the same name.


EXAMPLES

  To compress some files:

  > huffman -c foo.c bar.c test
  foo.c (38.2%)
  bar.c (33.1%)
  test (27.6%)

  To uncompress them:

  > huffman -u foo.c bar.c test

  To estimate the compression ratio:

  > huffman -r foo.c bar.c test
  foo.c (38.2%)
  bar.c (33.1%)
  test (27.6%)


BUGS

  Does not work perfectly under WB 2.0, v36.209.

  Option 'f' (filter mode) not implemented yet. Note that due to nature of
the huffman algorithm, it is necessary to read the input file two times,
one to determine the codification, and one to really compress. This makes
necessary to buffer the standard input to be able to re-read it later.

  Although the program does compress both text and binary files, and
does not seem to fail in any of the tests made with it, it has not been
tested in extreme conditions. Since the program destroys the original file
when compressing or decompressing, care should be taken.


AUTHORS

  The  accompanying  program  and  this documentation were written by Bruno
Costa and Lucia Darsa.  They can be reached in the following ways:

  E-Mail: bruno@brlncc.bitnet (InterNet)
          bruno@BRLNCC        (BitNet)

  Mail:   Bruno Costa
          Rua Almirante Guilhem 454/102 bloco 1
          Leblon - Rio de Janeiro
          RJ 22440
          BRAZIL


COPYRIGHT

  This  program  is  copyrighted by Bruno Costa and Lucia Darsa.  It may be
freely  distributed  if  and only if the source code, the documentation and
the  executables  are  provided  in  original  form.  It may be included in
Public  Domain/Shareware compilations, like the one organized by Fred Fish,
as  long as just a nominal fee is charged for each disk (no more than US$10
per  disk),  and may also be uploaded to BBS's or network services like BIX
and CompuServe.


DISCLAIMER

  The authors make no warranty of any kind with respect to this product and
explicitly  disclaim  any  implied warranties of merchantability or fitness
for any particular purpose.