Multi-Langauge Support

[agavra]

We should implement the Imprint spec in multiple languages so that it can be used in various different projects. Ideally we have at least a native Java & C implementation and use language bindings into Rust from Python/Typescript.

[expanded-for-real]

Are you looking for a full Java implementation of the spec (as in full gradle project) or something closer to an interop/compatibility layer?

[agavra]

@expanded-for-real for Java I think we should have a full native project instead of an interop layer so we can avoid shifting data between the JVM and native memory space. For now there's likely going to be a few rapid changes to the data format, so it may make sense to hold off on implementing the dedicated Java implementation until it's a bit more stable.

[expanded-for-real]

@agavra understood re:changing implementation. I was able to get a java version going based on the current implementation by just copying the Rust version and making use of Java's ByteBuffer to try to maintain zero-copy. It's currently in my forked repo - you can use some, all, or none of it. It's fairly simple so as long as the implementation doesn't get too out of hand then it should be (hopefully) easy to update.

[agavra]

@expanded-for-real I quickly took a look at the code and this is awesome! I love the JMH benchmark suite too.

I'd be happy to port this over into a new repo imprint-serde/imprintj or something like that. If you're open to it I can make you a committer for that new repo. I can try it out with Kafka Streams + Java.

[agavra]

https://github.com/imprint-serde/imprint-java - I created the repository and added you as a collaborator 🚀 happy to have you onboard, I can review the code in more depth if you want to issue the initial PR against imprint-java