Open saig0 opened 3 years ago
This would mean I need to implement the deserialization in other languages if I want to implement a new hazelcast client right?
This would mean I need to implement the deserialization in other languages if I want to implement a new hazelcast client right?
If we drop the support for Protobuf then yes. A new client would need to write the POJOs by itself. It could not use the Protobuf schema to generate the POJOs.
This is a good point for keeping the Protobuf support. On the other side, the Protobuf schema must be maintained to keep it in sync.
Since version
1.2.0
, Zeebe supports deserialization of its records from JSON to the Java objects. See here: https://github.com/camunda-cloud/zeebe/tree/develop/protocol-jacksonThe Java client should support the JSON format and deserialize the records using the new module.
The current client deserializes the records into objects that are generated from the Protobuf schema. Since these objects are different, we could create a new client. In the long run, we might support only the JSON format with the official module.