[
https://issues.apache.org/jira/browse/FLINK-25702?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jing Ge updated FLINK-25702:
----------------------------
Affects Version/s: 1.14.3
> Use the configure feature provided by the kafka Serializer/Deserializer.
> ------------------------------------------------------------------------
>
> Key: FLINK-25702
> URL: https://issues.apache.org/jira/browse/FLINK-25702
> Project: Flink
> Issue Type: Improvement
> Affects Versions: 1.14.3
> Reporter: Jing Ge
> Assignee: Jing Ge
> Priority: Major
> Labels: pull-request-available
>
> It is hard to understand the reason of using {{Configurable}} for
> KafkaRecordDeserializationSchema, since Deserializer already has the similar
> method for configuring.
> It assumes that users might have defined a deserializerA as a subclass of
> both Configurable and Deserializer, where
> (Configurable){{{}deserializerA::configure(Map<String, ?>){}}} could have
> user-defined configuration logic. Since this subclass could be defined by
> user outside the Flink project, we want to continue to support this this
> user-defined deserializer for backward compatibility.
> This backward compatibility has been considered. This idea is to just reduce
> the type control of the input parameter DeserializationSchema a little bit
> from {{Configurable & Deserializer<V>}} to {{{}Deserializer<V>{}}}. The
> wrapper will check internally and call
> {{org.apache.kafka.common.Configurable.configure}} if the schema implements
> {{{}Configurable{}}}. In this way, old cases due to the backward
> compatibility will be supported implicitly without confusing users who use
> {{Deserializer.configure}}
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
