Dandandan commented on PR #9659:
URL: https://github.com/apache/arrow-rs/pull/9659#issuecomment-4187404147

   > > For clickbench I found it is often best to just use the first or first 
two row filters, the other ones don't seem to bring much extra.
   > 
   > I think we can handle that on the datafusion side, what we have here (more 
efficient handling of the filters that do get pushed down) is good. We'll end 
up with more coarse control of IO / CPU patterns on the datafusion side and 
more granular and dynamic adaptivity in arrow-rs.
   > 
   > I do like that `FilterSelectivityStat` is exposed - I could see that being 
useful for adaptivity on the datafusion side.
   
   I agree with that - although I wonder what is the minimum amount of levers 
we need 🤔 


-- 
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]

Reply via email to