mbutrovich commented on PR #20806:
URL: https://github.com/apache/datafusion/pull/20806#issuecomment-4031576530

   > My only other thought would be to think about the end game on how this 
will be most maintainable.
   > 
   > Would this be easiest to maintain if it were _only_ a separate `Stream` 
implementation, with an ExecutionPlan shared with `SortMergeJoin`? We are 
downstream using a [physical optimizer rule to force `SortMergeJoin` to be used 
on sorted 
inputs](https://github.com/paradedb/paradedb/blob/f96213bf7675fbb81936fb52a339d62918f05761/pg_search/src/postgres/customscan/joinscan/planner.rs#L44-L68),
 and having only one "SortMergeJoin" operator (even if it had multiple backing 
implementations) would be a bit easier for us.
   
   I didn't look too closely at the rule, but I wonder if you could look for 
joins that both have `required_input_ordering()` and `maintains_input_order()` 
rather than the single `SortMergeJoin` operator. That said, the `Stream` 
implementation is an interesting idea and I'll see what that would look like.


-- 
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