The StrawTies Class

The two StrawTies objects provide a one-to-many mapping between keys (labels), and other Haystack objects. This essentially provides us with a method to maintain the list of ``edges'' that any vertex is attached to. StrawTies therefore represent the function f, where f(s) = {TieWeavable objects}, where s is a string label.

StrawTies are internally implemented as two Hashtables. One table maps a key to set of pointers (in the Java sense), and the other contains the mapping between a key and the HaystackIDs (see below) of the objects. In order to save the HDM structure persistently we serialize (see Section ) each Straw object to save on disk. Because the Java serialization mechanism will save not only the object being serialized but any objects that the serialized object points at, it is necessary to break apart Straws before saving them. Maintaining the two hashtables is an optimization that allows us to break apart and reassemble Straws easily.

We enforce the following invariant: let each vertex of the HDM graph (the Straw) possess a function, f₁ that accepts as input some key, k and produce a set of connected edges, A, labeled k. Let f₂ map some key j to a set of identifiers, B. Then " keys k, j, where k = j, $ a in A and b in B, s.t. a.getID() = b. Simply stated, the hashtable holding HaystackIDs as values can be used to produce the hashtable holding Straws as values.

StrawTies provide a subset of the features of the Java Hashtable. The function getVal(label) returns the Vector of values with the given label. To place a new element into the table, put(key,value) is used. Alternatively, remove(...) and removeKey(...) allow for pointers (i.e. edges) to be removed. getAllLinks(), getIDs(), isEmpty(), and toString() perform obvious functions.