circe VS ScalaPB

Using Circe so I define some classes that contain my custom Encoder[BusinessObject] in a file and I use that whenever I want to save/store a record, or handle a web request or respose. I also represent my mongo queries as JSON objects that I can freely build then pass to the driver.

Circe adopters should be using Scala https://github.com/circe/circe

If you look at Circe's github repo you will see a very large list of very recognizable companies, that should give you some idea. Circe isn't the ONLY Json parsing library, but it is probably the most popular, so - should give you a rough idea of the types and variety of companies using Scala.

Has the worst error messages I've ever seen for a parser. "Attempt to decode value on failed cursor" is not helpful when all you have is missing fields. Has been an issue for 5 years.

Circe tries to drop Scala 2.12 support in retaliation for not enough users paying them.

Circe: https://circe.github.io/circe/

PR for the same functionality in Scala3: https://github.com/circe/circe/pull/1800

Otherwise I tend to just use ZIO-JSON or Circe both of which have been updated for Scala 3.

I've updated results of benchmarks of 12 JSON parsers for Scala: - AVSystem's scala-commons - Borer - Circe - DSL-JSON - Jackson - jsoniter-scala - Play-JSON, - play-json-jsoniter - Spray-JSON - uPickle - weePickle - zio-json

I've had great experience with ScalaPB and even better with fs2-grpc (which builds on top of ScalaPB).

One way I've solved this is to use Protobuf contracts to serialize json with ScalaPB. They have a way implementing oneof that keeps each object individual and separate. Even if you don't decide to use PB, you can take some inspiration from the implementation.

You could try out protocol buffers. Not sure how stable it is, but I noticed they have a Scala3 version of their runtime library. (Disclaimer: I have not tried it out myself)