[Meta] Logstash Javafication

[andrewvc]

Below is a prioritized list of projects to convert Logstash internals and plugin API interfaces into Java:

• Java Plugin API @jordansissel
• Java Input API
• Java Processor API (filters+outputs)
• Configuration API
• Runtime Plugin Loading Java API: Design extension loading for new Java plugins #8375
• Pipeline compile-to-java @original-brownbear
• Structured Logging in log4j2 @yaauie
• Java entrypoint @original-brownbear
• Queue multiple-write @colinsurprenant
  • Impacts Java Input API
• Port Rake jobs to Gradle
• Core specs to be executed from JUnit @original-brownbear
• (Porting core specs is out of scope for this)
• Java plugin publishing tooling (blocked on dependency: Java API)
• Setting YAML to Java
• Move to snake_yaml, at minimum
• Grammar (really low priority)

[original-brownbear]

@andrewvc Let's tick the "Pipeline compile-to-java" section here? As discussed the differential gain from going further here is kinda small. I guess we can make some incremental improvements to this every now and then, but big gains won't come of just improving the execution (that needs either moving Batch and the batch readers to Java or a stronger plugin API).
Java entry point is incoming in #8161 but I'm waiting for #8599 to be a little quicker when ironing out the last details.

[colinsurprenant]

@original-brownbear just to clarify, is the "Pipeline compile-to-java" also refers to/includes the bytecode generation?

[original-brownbear]

@colinsurprenant jup, the most performance critical parts of this are compiled at runtime now.

[colinsurprenant]

@original-brownbear ok thanks, so yeah I agree on your comment above. Once I am done on the PQ batch maximization I'll resume the multi-write work I started which will/should include include moving the batch readers in Java...

[kares]

sounds like all of this is done by now, could we close?