Release 0.2.0

[NEW] Can use user-provided implementation of sort algorithm (the default is java.utils.Collections.sort)
[NEW] Added support for user-defined storage formats
[NEW] Added support for running Cubert script as Azkaban job
[NEW] User defined operators can request to cache files and Rubix file indices
[NEW] Added DICTINCT shuffle command
[NEW] Added DICTIONARY shuffle command
[NEW] Added RANK operator
[NEW] Added support for CUBE combo level and rollup
[CHANGE] Inline script dictionary is now created using keyword CREATE DICTIONARY
[CHANGE] ENCODE and DECODE operators now require the columns that are to be encoded/decoded
[CHANGE] Package org.apache.pig.piggybank (for AVRO storage) is renamed to com.linkedin.cubert.pig.piggybank to avoid name conflicts
[PERFORMANCE] Caching classname (string) -> Class mapping to improve runtime performance
[PERFORMANCE] New implementation of hash table (for HASH-JOIN operator) that is memory efficient
[BUG] Fixed bug where cached file analyzer was incorrectly treating the inline dictionary as path to a file
[BUG] TEXT output format was ignoring the field separator value
[BUG] User defined aggregation functions with default constructors were not instantiated correctly
[BUG] Fixed issues where distributed cached files were not handled properly in hadoop 2

Release 0.1.0

First open source release of Cubert.