Collection

0.1 Rube Goldberg release

This library implements persistent datastructures for Rust, with efficient, with the complexity of the union operation can be sublinear in practical cases.

Deterministically balanced search trees

This is achieved by using deterministically balanced trees, where the split-points of the tree are determined by the Weight of the elements in the collection.

That T implements Weight, means that T, or parts of T can be Hashed this hash is then used to determine the weight, essentialy by counting leading zeroes in the hash. The weight then determines how many levels to split.

Copy on write, and structural sharing

Cloning a collection is a constant time operation. This is achieved by the Stash abstraction, which keeps a collection of Nodes, referenced by a Location indirection.

Pluggable metadata

Each collection can carry different kinds of metadata, metadata is defined as the operations &T -> Meta<T>, and a binary operation combining two Meta<T> with each other. So, to implement a Set, you would use Max<T>, and if you also want constant-time equality checking, you would add CheckSum<T>

Collection types

Collection comes with 3 pre-defined sets of operations making up a Set, a Vector, and a Map.

To define a collection, you use the collection! macro:

    collection!(Set<T> {
        max: Max<T>,
        checksum: CheckSum<u64>,
    } where T: Ord + Hash);

    collection!(Vector<T> {
        cardinality: Cardinality<usize>,
        checksum: CheckSum<u64>,
    } where T: Hash);

    collection!(Map<T> {
        key: Key<T::Key>,
        keysum: KeySum<u64>,
				valsum: ValSum<u64>,
    } where T: Keyed, 
            T::Key: Hash,
            T::Value: Hash);				

License

GPLv3