Introduce j4rs wrapper to simplify testing java drivers

[rukai]

The j4rs java interop crate is quite flexible but it allows for a few things which bring us no value:

• working with multiple JVM's
  • We only use a single JVM
• returns a Result<T, E> on every operation that touches the JVM in case the jvm explodes
  • In this case the only reasonable thing for us to do is panic.
• Differentiates between values returned from java Instance and values that will be sent into java InvocationArg
  • I believe this is done for performance reasons, but we dont need to be concerned about that level of performance since this is an integration test

All of these features were causing great complexity for our java driver integration tests, so I introduced a wrapper that does not expose the above features. It reduces the lines of code used to implement the kafka java driver from 687 to 484. ( 30% reduction in lines) We should expect a similar reduction in complexity for the cassandra java driver. The wrapper completely wraps j4rs and we should not interact directly with j4rs in our driver logic anymore.

The new wrapper consists of 2 types:

• Jvm - wraps j4rs::Jvm
  • constructor sets up JVM and installs any needed java dependencies from maven repository.
  • provides methods for creating new Values
• Value - wraps j4rs::Instance and additionally takes on the role of j4rs::InvocationArg
  • provides methods for interacting with the java value e.g. call methods or access fields

While this wrapper is closely matched to our own needs and probably not upstreamable exactly as is. We might be able to see parts of it upstreamed either as an optional wrapper or integrated with the main API. Doing so would reduce the complexity of the wrappers implementation. But no hurry, we can land the wrapper as is for now.

[codspeed-hq[bot]]

CodSpeed Performance Report

Merging #1682 will improve performances by 13.12%

_{Comparing rukai:java_interrop_wrapper (44d5124) with main (f53bf73)}

Summary

⚡ 1 improvements ✅ 38 untouched benchmarks

Benchmarks breakdown

	Benchmark	main	rukai:java_interrop_wrapper	Change
⚡	encode_system.local_result_v5_no_compression	105.5 µs	93.3 µs	+13.12%