performance regression (?) when compiling a Spring Boot application with GraalVM

[joshlong]

i've noticed that in Spring AI m3 my graalvm native image starts almost twice as slowly as it did in m2, despite there not be any other observable changes.

here's an example: https://github.com/joshlong-attic/spring-ai-m3-graalvm-regression

to see it in action:

• start Docker
• run: init_db.sh
• compile the mvn Spring Boot application in the usual way: ./mvnw -DskipTests -Pnative native:compile
• start it: ./target/service