optimize startup of java-function-invoker about riff HOT 5 CLOSED

Shall we close this, or do we want to push further?

from riff.

We should document what we did and measure it, so we can assess regressions or improvements. I made some benchmarks. Will post results soon.

For the record, classpath scanning is (really) never the problem. Perpetuating the myth is a bad idea.

from riff.

I agree we should document our benchmarks as a reference. I'd like to know if @dsyer thinks there's more to do in the short term as well.

re: classpath scanning - that's good to know, Dave.

from riff.

#139 is probably as good as it gets for the short term at least. Benchmarks here: https://github.com/dsyer/java-function-benchmarks. TL;DR we can implement Flux functions without affecting performance, but possibly not with full dependency injection and Boot features.

from riff.

cool. nice work @dsyer !

from riff.