Github user kevinjmh commented on the issue:
https://github.com/apache/carbondata/pull/2767
Hi @ravipesala , I have some question about this PR
1. It is a little confusing to only deal with AND case and ignoring OR case
if we want the framework do expression merging in general.
2. This PR still failed when querying on STRING column with bloom filter
like "column1 = 123". Please refer to case 1 in description of PR #2665 for
details
3. One more fail case found if apply this PR:
3.1 query with filter alike `(Col1='a' or Col2 ='b') and (Col3='c' or
Col4='d' ) `
3.2 ensure test result without bloom should not be empty
3.3 create bloom filter on multi columns in same datamap
3.4 test result with bloom would be empty. Because OrExpression is not
proceeded when creating bloom query model now.
Please check it.
One more thing I think we could do to improve datamap pruning efficiency is
to reuse pruning result between filter conditions. Such as we query with two
filter conditions on segment with 5 blocklets, we get 2 hit blocklets after
applying first filter condition, and then the second filter condition can only
apply on those 2 blocklets instead of 5.
---
