kjems / ComposedSet

A collection that decomposes it's elements into smaller chunks which is then stored in an array. Each element is then represented by indices into the array of parts. Many operation on the collection can be done by looking at the indices.

ComposedSet

Use this library on a dataset when:

• Each data element in a dataset can be further split into parts
• Each part is typically repeated in other data elements
• There is not a constant need to compose back to the original data representation
• Performance on operations in the decomposed state matters
• Memory footprint is important and many of the parts are reused

Example : Folder structure

Assume we have the following paths

data/example/code/myfile.cpp
data/production/code/some.cpp

We will then split using '/', '.' and put each unique part in a indexable collection :

[0:data | 1:/ | 2:example | 3:code | 4:myfile | 5:. | 6:cpp | 7:production | 8:some]

The two paths can now be represented as indicies into the shared collection of parts:

[0,1,2,1,3,1,4,5,6]
[0,1,7,1,3,1,8,5,6]

Now operation like 'find all .cpp files in the data folder' becomes a lot faster than comparing the raw strings because that will just be checking if the indicies they start with [0,1] (data/) and end with [5,6] (.cpp).

Concatination and substring operations also become significantly faster.

Build Status

Mono	.NET

Maintainer(s)

• @kjems