Java Minimalist & Modern Stack

Minimalist stack for Java

Java Modern Stack

• Runtime Environment: JDK 21 (LTS)
  • Chosen over older versions (such as JDK 17 or 11) to leverage Virtual Threads (Project Loom) for high-performance concurrent request handling, along with pattern matching for switch, record patterns, and sequenced collections.
• Build Tool & Package Manager: Gradle with Kotlin DSL
  • Chosen over Maven due to Gradle’s superior build speed (using incremental compilation, build cache, and daemon), and the type-safety, auto-completion, and readability of writing build scripts in Kotlin rather than verbose, rigid Maven XML.
• Linter & Formatter: Spotless with google-java-format
  • Chosen over Checkstyle / PMD because Spotless actively formats code automatically at compile or commit time using Google’s official style guide, rather than just raising static analysis warnings that require manual developer intervention.
• Test Runner: JUnit 5
  • Chosen over JUnit 4 or TestNG because of its modular Jupiter platform architecture, rich extensibility, nested test suites, parameterised testing, and native support for modern Java language features and parallel execution.
• Web/Application Framework: Spring Boot or Quarkus
  • Spring Boot is recommended for standard enterprise apps due to its unmatched ecosystem size, mature security frameworks, and deep library support.
  • Quarkus is recommended for cloud-native/minimal microservices due to its compile-to-native GraalVM support, sub-second startup times, and tiny memory footprint.
• Boilerplate Reduction: Lombok
  • Chosen over manual boilerplate writing because it uses annotations to generate getters, setters, equals/hashCode, toString, and builder patterns at compile time, improving readability and reducing code size.
• Database Client: jOOQ
  • Chosen over Hibernate / JPA because jOOQ is a SQL-centric, type-safe query builder that generates Java classes directly from database schemas. It gives developers full control over SQL syntax and query optimization while avoiding the runtime magic, N+1 query traps, and cache synchronization problems of full ORMs.

Edit on GitHub · Back to index