[ https://issues.apache.org/jira/browse/FLINK-2026?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14550017#comment-14550017 ]
Sebastian Schelter commented on FLINK-2026: ------------------------------------------- Wouldn't it be best to simply ignore the call to execute() if there is no more work to do? I as user don't want to think about all the internals to be honest. For me its common to do add a env.execute() call at the end of my flink code. Always having to remove that if I use a count or print call somewhere seems tedious. > Error message in count() only jobs > ---------------------------------- > > Key: FLINK-2026 > URL: https://issues.apache.org/jira/browse/FLINK-2026 > Project: Flink > Issue Type: Bug > Components: Core > Reporter: Sebastian Schelter > Assignee: Maximilian Michels > Priority: Minor > > If I run a job that only calls count() on a dataset (which is a valid data > flow IMHO), Flink executes the job but complains that no sinks are defined. -- This message was sent by Atlassian JIRA (v6.3.4#6332)