|Maintainer||Timo B. Huebel (firstname.lastname@example.org)|
Version : 0.1
This module provides compression for streams of 32-bit words. Because of some internal restriction in GHC, which makes all fixed integer size equal in terms of bit-width, the algorithm tries to crunch as much numbers as possible into a single 64-bit word.
Based on the Simple9 encoding scheme from this article:
- Vo N. Anh, Alstair Moffat, "Inverted Index Compression Using Word-Aligned Binary Codes", Information Retrieval, 8 (1), 2005, pages 151-166
Crunch some values by encoding several values into one
Word64. The values may not exceed
the upper limit of
(2 ^ 60) - 1. This precondition is not checked! The compression works
best on small values, therefore a difference encoding (like the one in
Holumbus.Data.DiffList) prior to compression pays off well.