ercsonusharma commented on PR #4546:
URL: https://github.com/apache/solr/pull/4546#issuecomment-4795414752

   > Can't we use the existing 
org.apache.solr.common.params.ShardParams#SHARDS_ROWS for this use-case?
   I thought about this but since it was specific to rrf and have to put some 
limit, I chose this. But since, we don't need the limit so we can re-use that 
param with a minor fix at 
[mergeIds](https://github.com/apache/solr/pull/4546/changes#diff-1ca21c13d967412ea6227fe257616a9cb2ac45012b207373a526d3767a46b588R361).
   
   > I can't tell but would a hypothetical queryResultWindowSize of say 20 mean 
that a page (rows) of 10 on first & second pages (start=0 & start=10) should 
get consistent results, even without this PR?
   
   The inconsistency in combined query paging happens at the coordinator's RRF 
step, not the shard query step. For above example, `queryResultWindowSize=20` 
would just mean each shard's queryResultCache happens to already hold 20 docs, 
so the page-2 shard request is a cache hit instead of a re-execution. But the 
coordinator still receives 10 candidates on page 1 and 20 on page 2, and RRF 
over a 10-doc pool for page 1 produces different scores/ordering than RRF over 
a 20-doc pool.


-- 
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