[ https://issues.apache.org/jira/browse/CALCITE-2970?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17095895#comment-17095895 ]
Xiening Dai commented on CALCITE-2970: -------------------------------------- Yes, the #1 point I mentioned. If there's a sub-class RelBuilder, we won't be able to handle. I agree it's not a great practice for sub-classing, but the reality is it has been there with Flink, MaxCompute, and others. Also for #2 it's actually quite common in federation - you have an XTableScan for heterogeneous data source, then use build-in EnumerableFilter and EnumerableJoin to process. Also today RelOptRuleCall already has a RelBuilder object, which is used by most of the rules. It would be a waste if we discard that and create a new one. > Performance issue when enabling abstract converter for EnumerableConvention > --------------------------------------------------------------------------- > > Key: CALCITE-2970 > URL: https://issues.apache.org/jira/browse/CALCITE-2970 > Project: Calcite > Issue Type: Bug > Components: core > Reporter: Haisheng Yuan > Priority: Major > Labels: pull-request-available > Time Spent: 17h 10m > Remaining Estimate: 0h > > If we enable the use of abstract converter for {{EnumerableConvention}}, by > making {{useAbstractConvertersForConversion}} return true, > {{JDBCTest.testJoinManyWay}} will not complete. -- This message was sent by Atlassian Jira (v8.3.4#803005)