[ https://issues.apache.org/jira/browse/FLINK-8437?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Chesnay Schepler updated FLINK-8437: ------------------------------------ Description: The current API for retrieving Side Outputs is a bit ambiguous. Consider the program below: {code:java} sideOutput = stream .process(...) .filter(...) .getSideOutput(tag) {code} This may be the sideOutput of the process function that is passed through the API for convenience, or the sideOutput of the filter function (which would always be empty). Given that only process functions can have sideOutputs we may want to change the return type so that getSideOutput can only be called after a process function. was: The current API for retrieving Side Outputs is a bit ambiguous. Consider the program below: {code:java} sideOutput = stream .process(...) .filter(...) .getSideOutput(tag) {code} This may be the sideOutput of the process function that is passed through the API for convenience, or the sideOutput of the filter function (which would always be empty). > SideOutput() API is ambiguous > ----------------------------- > > Key: FLINK-8437 > URL: https://issues.apache.org/jira/browse/FLINK-8437 > Project: Flink > Issue Type: Improvement > Components: Streaming > Affects Versions: 1.5.0 > Reporter: Chesnay Schepler > Priority: Minor > > The current API for retrieving Side Outputs is a bit ambiguous. Consider the > program below: > {code:java} > sideOutput = stream > .process(...) > .filter(...) > .getSideOutput(tag) > {code} > This may be the sideOutput of the process function that is passed through the > API for convenience, or the sideOutput of the filter function (which would > always be empty). > Given that only process functions can have sideOutputs we may want to change > the return type so that getSideOutput can only be called after a process > function. -- This message was sent by Atlassian JIRA (v7.6.3#76005)