berkaysynnada commented on PR #20138:
URL: https://github.com/apache/datafusion/pull/20138#issuecomment-3843984257

   The fix seems correct to me. To elaborate, assume we have 2 child intervals; 
A and B. Propagation of TRUE over Eq is done like that:
   
   We prune A such: "choose a point in A. If that point cannot be satisfied by 
any point from B, then that point is pruned."
   and for B, similarly: "choose a point in B. If that point cannot be 
satisfied by any point from A, then that point is pruned."
   
   So, what I think now is, we don't actually require interval sets for this 
problem, only except this case:
   
   A is a set which strictly contains B, and B is a single-point value. In that 
case, we need to split the interval A into 2 parts at the intersection point 
(or vice versa for B and A). However, leaving the fix as is is not a big 
problem. We will only not be capable of a tiny optimization.
   
   But to not introduce another bug, the only case which should keep as 
infeasible is, having 2 same single-point values. In that case, having an 
equality as false is impossible.
   
   cc @ozankabak if I missed anything


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to