Group by Pipeline/Group by Hash
I have a query which does a union between 2 datasets.
The underlying projections on both the tables are ordered by columns a,b,c to get "Group by Pipeline". It works perfectly fine when both tables have data.
But per a valid business scenario, If one of the tables is empty(say abc in below example) it opts for "Group by Hash" and query dies down. Please suggest.
select a,b,c from xyz group by 1,2,3
select a,b,c from abc group by 1,2,3