Abstract base class usable by implementations that wish to share the same primary/secondary, lock-by-stripe approach.

Each bucket is a multi-record mini-hash table that stores multiple pointers into secondary. If a bucket overflows, we chain to a second bucket of poi

Object corresponding to a particular bucket.

This represents a chunk o' bytes that does not have a correspondence (yet) to a particular storage location.

Represents the result of a search in a bucket chain for a particular key. If the search was successful, we do not need to record the information abou

For now we make the assumption that bytewise equality is the way to check for serialized equality (as it is also the way we calculate hash codes for e

You should expect this to be pretty slow / inefficient, but it is a good fallback implementation.

An implementation that stores "buckets" consisting of single-record pointers, which are chained on collision in a secondary file where data is also st

This represents a bucket that exists at a particular location in an underlying file. Mutator methods will propagate through to the underlying storage