You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "Konstantin Knauf (Jira)" <ji...@apache.org> on 2021/07/29 07:38:15 UTC

[jira] [Updated] (FLINK-8378) Serialization and Deserialization of Kafka Avro messages from and to Row with Confluent Schema Registry

     [ https://issues.apache.org/jira/browse/FLINK-8378?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Konstantin Knauf updated FLINK-8378:
------------------------------------
      Labels:   (was: auto-closed)
    Priority: Not a Priority  (was: Minor)

> Serialization and Deserialization of Kafka Avro messages from and to Row with Confluent Schema Registry
> -------------------------------------------------------------------------------------------------------
>
>                 Key: FLINK-8378
>                 URL: https://issues.apache.org/jira/browse/FLINK-8378
>             Project: Flink
>          Issue Type: New Feature
>          Components: API / Type Serialization System
>    Affects Versions: 1.3.2, 1.4.0
>            Reporter: Christophe Philemotte
>            Priority: Not a Priority
>
> I need to serialize and deserialize Avro messages from Kafka whose the schema are stored in the Confluent Schema Registry to Row instances.
> As far as I know,
> * KafkaAvroTableSource does not fetch the schema from the registry
> * there is no TableSink for KafkaAvro
> * the AvroRowSerializationSchema and AvroRowDeserializationSchema needs a known Avro schema
> * there is none wrapper of the Kafka serdes
> So, for now it's not possible out of the box. Still, I think it's an interesting use case especially considering the Table & SQL API.
> I've found some issues related to my case:
> * https://issues.apache.org/jira/browse/FLINK-4050
> * https://issues.apache.org/jira/browse/FLINK-2597
> It seems to me that they have wider scope, that's why I'm creating this issue. I hope I'm doing well.
> I've managed to implement a serializer and deserializer that are between the Kafka ones and the Flink AvroRow ones.
> If you think such serdes would be a nice addition, I'm ready to contribute and rework my implementation (notably to work with TypeSerializers if possible).



--
This message was sent by Atlassian Jira
(v8.3.4#803005)