The Java Platform continues to change and one of the changes which will likely happen in the next couple of years (e.g. by 2026), is Project Leyden. At the level of the Java platform the idea is to improve things like startup time performance by making use of "link time" optimizations (i.e. after compile time, but before runtime).

Kafka application developers themselves will likely want to make use of these new features of the Java Platform. But from a Red Hat PoV Project Leyden will be a very big deal for Quarkus.

In order to work Leyden needs to constrain the Java application (this is similar to, but weaker than Quarkus's "closed world" constraint). In other words, libraries won't be able to use certain features of the Java Platform, including limitations on things like reflection and native code. It's also likely that libraries will need to be JPMS compatible. I.e. that you can run them using the module path rather than the classpath.

The Kafka client libraries currently use reflection to instantiate plugins such as serializers and deserializers. They also rely on native code in the implementation of compression codecs.

This Epic is to cover work that will be needed for Kafka clients to work with Leyden, including investigating in detail what's actually required.