Dandandan commented on code in PR #3578:
URL: https://github.com/apache/arrow-datafusion/pull/3578#discussion_r996295305
##########
benchmarks/expected-plans/q7.txt:
##########
@@ -14,7 +14,9 @@ Sort: shipping.supp_nation ASC NULLS LAST,
shipping.cust_nation ASC NULLS LAST,
TableScan: lineitem projection=[l_orderkey, l_suppkey,
l_extendedprice, l_discount, l_shipdate]
TableScan: orders projection=[o_orderkey, o_custkey]
TableScan: customer projection=[c_custkey, c_nationkey]
- SubqueryAlias: n1
+ Filter: n1.n_name = Utf8("FRANCE") OR n1.n_name =
Utf8("GERMANY")
Review Comment:
Think so too - although if it stays on the probe side (e.g. the
HashBuildProbeOrder rule doesn't apply) - the joins might still quite a bit
slower, and the input and output of the joins will be bigger as well making
them more expensive.
FYI @andygrove this might change the perf in Ballista too.
--
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]
