Author: Stan Eisenstat
Subject: Re: [Cs323] Clarifications
Date: Monday, 26 Oct 2020, 20:16:19

    > Message Posted By: Unknown
    >
    > Since this is known: "D. When encode writes a code, the number of bits
    > depends on the number of codes assigned, not the value of the code.
    > decode knows this number, so it can deduce how many bits encode sent", why
    > do we need to do send a code for GROW?

Sending a code may be easier than deducing.
=====

    > Also, depending on the number of special codes stored in the table (say
    > 3), wouldn't that affect the actual degree to compression in some cases.
    > For example, if there is one case with -p window option where when you
    > have 2 special codes and you reach MAX_CAPCITY -1 whereas in 3 special
    > codes you reach MAXCAPCITY for table and then prune (which leads to more
    > memory). Do we have to think about these type of cases (for the 9*SIZE
    > constraint) or will the test script automatically adjust for these.

The size of the output from encode may change slightly
as more special codes are reserved.  Thus the test
scripts will allow some variation when comparing the
sizes of the output from your encode with that from
Hwk4/encode.

--Stan-