tree 03fe751c8c262e3e47c033a8803dc138c52a1497
parent b0bdab2597dd658eceaea0b87672c76e06eaf340
author QiangCai <qiangcai@qq.com> 1577447787 +0800
committer Jacky Li <jacky.likun@qq.com> 1577680217 +0800

[CARBONDATA-3641] Refactory data loading for partition table

[Background]

Currently, CarbonData only implemented hadoop commit algorithm version 1, which generated too many segment files during loading and generated too many small data files and index files

[Modification]
 1. implemented carbon commit algorithm, avoid to move data file and index files
 2. generate the final segment file directly
 3. optimize global_sort to avoid small files issue
 4. support complex data type in partition table (non-partition column)

This closes #3535