xuchuanyin created CARBONDATA-2944:
--------------------------------------
Summary: optimize compress/decompress related procedure
Key: CARBONDATA-2944
URL:
https://issues.apache.org/jira/browse/CARBONDATA-2944 Project: CarbonData
Issue Type: Sub-task
Reporter: xuchuanyin
Assignee: xuchuanyin
While implementing customize compressor, I found that carbon compressor deal with primitive object while compressing/decompressing which I think will cause efficiency because:
1. many compressor do not provide compress/decompress interface for primitive objects, we need to handle it ourselves, which may cause unnecessary conversion from primitive to bytes and from bytes to primitives;
2. for querying, we need to decompress the content. I think it's better to keep them in bytes and convert them to primitives until it is needed.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)