Just made the switch from IntelliJ Ultimate to VS Code for my main Java/Spring Boot project. The startup time difference is... noticeable. IntelliJ was taking a solid 45-60 seconds to become responsive on my M1 MacBook Pro (32GB RAM). VS Code, with the necessary extensions, pops open in under 10.
But I'm not just here for a speed boast. I'm trying to figure out if I've traded performance for essential functionality. My plugin list is growing, and I'm worried about conflicts creeping in.
**My current VS Code Java setup:**
- **Extensions Pack:** `vscjava.vscode-java-pack` (covers Language Support, Debugger, Test Runner, Maven/Gradle, Project Manager)
- **Spring Boot:** `vmware.vscode-spring-boot`
- **Lombok:** `vmware.vscode-lombok`
- **Tomcat/Jetty:** `redhat.vscode-server-connector`
- **Checkstyle:** `shengchen.vscode-checkstyle`
- **SonarLint:** `sonarsource.sonarlint-vscode`
**Observations & Potential Conflicts:**
* The Java pack's language server (`jdt.ls`) seems robust, but I occasionally get duplicate diagnostics from SonarLint and Checkstyle on the same issue. Is there a way to deduplicate or prioritize?
* I've heard the Spring Boot and Lombok extensions can sometimes fight over annotation processing. So far, so good, but I'm watching for strange symbol resolution errors.
* Memory usage seems lower overall, but I'm curious if anyone has benchmarked `jdt.ls` memory consumption with a large multi-module Maven project (>200 modules).
Has anyone else run this migration gauntlet? I'm specifically looking for:
- Known performance hits or conflicts within this extension combo.
- Any essential `.vscode/settings.json` tweaks for Java that prevent slowdowns.
- Whether adding AI assistants (Tabnine, Cody, Cursor-like features) on top of this stack has caused issues for you.
I'll run some side-by-side code completion latency benchmarks next week and post the results. For now, the startup win feels great, but I'm wary of the long-term health of this setup.
--experiment
Prompt engineering is the new debugging.