6.826 Problem Set 2 Clarifications

Problem 1.

The intention of grouping "code" and "decode" functions in
the module header is to state the invariant that they are
inverse of one another.  You can use the code and decode
functions by simply writing code(b) and decode(b) where b is
a Byte.

When determining whether CypherFile implements File and
whether File implements CypherFile, consider only traces
that contain Read and WriteAtomic operations, do not worry
about the fact that File has some additional operations.
(You still need to look at the implementation of NewFile
because it is used to implement WriteAtomic.)

Recall that the action corresponding to a function call or
atomic procedure call is labelled by both the parameters and
the return result of the function or procedure.

When analyzing the effect of NewFile operation you may
assume that for every call WriteAtomic(pn,x,data) the write
does not start after the end of file: it is always the case
that x <= d(pn).size

Problem 2.

d) You should consider the set of states that can be
generated by Read and Append operations in a file system
with existing EMPTY files.  The point of this requirement is
that the state is nontrivial, but is still generated only by
Read and Append operations.  You assume in the d) part that
the invariant in b) holds for the state.

Problem 3.

You will need to make modifications to the state to maintain
the additional information which will allow you to implement
the cache replacement policies.  Write all modifications
that you make to the original implementation.