Github user winningsix commented on the issue:
https://github.com/apache/orc/pull/189
@omalley Thank you for the patch and initial performance result. The
result looks very promising. With current result, looks like different encoding
(FPC V1, FPC V2 and split) fits for different data set while split encoding is
more compression friendly. Not sure whether we can implement it as what RLE
did. Dynamically determine the encoding based on the characteristics of data
set. Any thoughts on this?---
