Github user tzulitai commented on a diff in the pull request:
https://github.com/apache/flink/pull/5958#discussion_r186338721
--- Diff:
flink-core/src/main/java/org/apache/flink/api/common/serialization/DeserializationSchema.java
---
@@ -42,14 +42,22 @@
@Public
public interface DeserializationSchema<T> extends Serializable,
ResultTypeQueryable<T> {
+ /**
+ * @deprecated Use {@link #deserialize(ConsumerRecordMetaInfo)} .
+ */
+ @Deprecated
+ T deserialize(byte[] message) throws IOException;
+
/**
* Deserializes the byte message.
*
- * @param message The message, as a byte array.
+ * @param consumerRecordMetaInfossage The message, as a {@link
ConsumerRecordMetaInfo}.
*
* @return The deserialized message as an object (null if the message
cannot be deserialized).
*/
- T deserialize(byte[] message) throws IOException;
+ default T deserialize(ConsumerRecordMetaInfo
consumerRecordMetaInfossage) throws IOException {
--- End diff --
I'm actually not sure that we should continue using this class, for the
following reasons:
1. The class is actually placed under a non-ideal package:
`o.a.f.api.common.serialization`, whereas is should be placed under some
`o.a.f.connectors.kafka....`.
The reason it is currently placed under this package was because the
`DeserializationSchema` was initially intended to be commonly used by all
connectors. However, over time, things have proven that each connector will
benefit from their own version of a schema class.
So, it actually might make sense to deprecate the whole
`DeserializationSchema` class now, and have a new class (maybe called
`KafkaDeserializationSchema` / `KafkaSerializationSchema`) under a correct
Kafka package.
What do you think?
---
