Claus Ibsen created CAMEL-26121:
-----------------------------------
Summary: camel-flink - the component dataStream and
dataStreamCallback options are ignored, and a DataStream callback always fails
with the default collect=true
Key: CAMEL-26121
URL: https://issues.apache.org/jira/browse/CAMEL-26121
Project: Camel
Issue Type: Bug
Components: camel-flink
Reporter: Claus Ibsen
Fix For: 4.24.0
Found while adding the catalog body types of flink, spring-batch and
optaplanner (CAMEL-24961, PR #28418).
h4. 1. The component-level dataStream and dataStreamCallback options are ignored
{{FlinkEndpoint.doInit}} (main 2a465aa4814c, lines 74-84) falls back to the
component's {{dataSet}} and {{dataSetCallback}}, but not to its {{dataStream}}
and {{dataStreamCallback}}. With those set only on the component,
{{flink:datastream}} fails "No DataStream defined"
({{DataStreamFlinkProducer}}, line 95). Checked on 4.24.0-SNAPSHOT with a local
Flink environment.
Proposed: also fall back to the component's {{dataStream}} and
{{dataStreamCallback}} in {{doInit}}.
h4. 2. A DataStream callback always fails with the default collect=true
{{collect}} defaults to {{true}} ({{FlinkEndpoint}}, main 2a465aa4814c, line
50), and {{DataStreamFlinkProducer.collectResults}} (line 79) throws "collect
mode not supported for Flink DataStreams." when the callback returns a
DataStream. So a DataStream callback returning a DataStream fails until
{{collect=false}} is set. Checked on 4.24.0-SNAPSHOT with a local Flink
environment.
Proposed: for {{flink:datastream}} return the DataStream as is when {{collect}}
was not set explicitly (or default collect to false there).
_Claude Code on behalf of davsclaus_
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
