A toolkit for dealing with data on the wire.
Dealing with schema evolution is a pain. Kopper aims to make it easier by providing a set of tools to help you manage your schema evolution and decouple how data is used from how it is stored.
Kopper core introduces a new way of working with data in your application. Instead of generating out code, it takes the approach of writing an interface and seamlessly link it to the data underneath.
Kopper Core currently has a big focus on Avro, since that's what we are using at KOR. Generating out SpecificRecord classes causes compatibility issues that aren't easy to solve. But GenericRecord is a pain to work with because it simply is too generic. We actually want to have a strong-typed interface to our data.
Dynamic Proxies allow us to define an interface and define what needs to happen when a method is called. This is exactly what we want; the strong typed interface translating to and from a GenericRecord.
Serializers and Deserializers are available to hook these dynamic records into your kafka streaming logic.
Kopper Cast takes schema evolution to the next level by defining what needs to happen to evolve from one schema version to another. These so-called 'casters' can then be used to read and write data in a compatible way, without the restrictions resulting from using compatibility modes.
Kopper cast allows for schemas evolving in an incompatible way, having the casters handle the incompatibility.
Kopper is not necessarily production-ready code (yet). It lays out ways to implement conceptual patterns around schema evolution. Kopper is open to community contributions and interactions, and can over time become hardened enough to qualify for production use cases.