Author: Stan Eisenstat
Subject: Re: [Cs323] pruning
Date: Sunday, 01 Nov 2020, 08:12:01

    > Message Posted By: Unknown
    >
    > How much shorter does pruning have to be on public test cases? Is the
    > pruned amount compared to the original file size or a specific non-pruned
    > encoded output size? Is there a certain percentage that we can expect for
    > how much shorter it should be using "wc -c FILE"

As stated in the specification:

  L. The degree of LZW compression (i.e., the length of the output from encode)
     depends on the file; the value of MAXBITS (and how this value is represented
     in the output from encode); and the number of special codes (e.g., EMPTY,
     GROW, or PRUNE).  Thus all tests of size will be in comparison with that
     given by Hwk4/encode (which the scripts assume is a correct implementation
     of LZW) and will be relatively loose (at least 1%).

--Stan-