Hi folks, does anyone know what's the order in which indices are checked for a query? Say I have a query like the following:

select * from table where col1 >= 10 and col2 = 'foo' and col3 = 'bar' and col4 < 20

Assuming col1, col4 have range indices and col2, col3 have inverted indices. Would the indices be picked based on the order of the filter conditions?

Good question. Looking at this code, I don’t see a tiebreaker: https://github.com/apache/pinot/blob/64352999e439467431116b32d2e69be7469c768a/pino[…]/java/org/apache/pinot/core/operator/docidsets/AndDocIdSet.java

It doesn’t really matter because we will evaluate all of them

But for

and

don’t we evaluate in prioritized manner so that the output of more efficient filtering is fed into the next predicate?

For index backed by bitmap, we will evaluate them and do and operation on all bitmaps

the range index could support pushing bitmaps down in to the filter, which would make the range filters faster when there's a (selective) inverted index present