[ https://issues.apache.org/jira/browse/BEAM-3749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16380811#comment-16380811 ]
Kenneth Knowles commented on BEAM-3749: --------------------------------------- I realize a problem though - you do need to add the trigger after the read, so the way that IO is done in pure SQL CLI means you need to add the triggering after the IO, yet this is part of the query. So actually we might need a change like this one you suggest, but to set up triggering on the table scan relnodes. > support customized trigger/accumulationMode in BeamSql > ------------------------------------------------------ > > Key: BEAM-3749 > URL: https://issues.apache.org/jira/browse/BEAM-3749 > Project: Beam > Issue Type: Improvement > Components: dsl-sql > Reporter: Xu Mingmin > Assignee: Xu Mingmin > Priority: Major > Fix For: 2.4.0 > > > Currently BeamSql use {{DefaultTrigger}} for aggregation operations. > By adding two options {{withTrigger(Trigger)}} and > {{withAccumulationMode(AccumulationMode)}}, developers can specify their own > aggregation strategies with BeamSql. > [~xumingming] [~kedin] [~kenn] for any comments. -- This message was sent by Atlassian JIRA (v7.6.3#76005)