A combined compress(1) and uuencode(1) implementation for bash/ksh
LZW encoding is a variable bit width dictionary lookup typically done with a
hash keyed by a couple (prev,cur). I use the filesystem instead to respresent
a trie, with the byte values as directory names, and a single file e to store
the dictionary index.
This is not very fast, and it's best to keep the COMPRESH_BITS setting low.
If you're patient (and the filesystem is ample), you can try setting it to the
compress(1) default of 16.
Enjoy!