Value Types and Pattern Matching: 1–1, Data at the Center
summary
Demonstrates concrete Java features and an LW5 prototype that implements value classes via an annotated bytecode rewriter and is benchmarked with JMH. Presents performance results: ~25 µs baseline for a primitive sum, ~70 µs when using boxed objects, ~30 µs with value records, and ~11 µs when arrays are flattened with non-null declarations (2x–6x speedups and much lower memory). Also shows record patterns and sealed types with exhaustive switch checking, and calls out practical gaps: switch specialization bugs, generic boxing without specialization, and the need for default-zero constructors plus atomic/non‑atomic semantics for flattening.
AI-written summary · about
chapters (6)
- 0:00Introduction
- 28:03Data-oriented programming
- 1:08:50Records and pattern matching
- 1:33:39Project Valhalla
- 2:33:23Valhalla and Amber integration
- 2:43:56Q&A
related talks
33mImprove your Java Applications with Records and Pattern Matching
29mThe Art of Java Language Pattern Matching
54mData Focused programming with Pattern Matching in Java
47mValue Objects - The Next Big Thing in Java
50mPattern Matching in Java Deep Dive
56mThe Art of Java Language Pattern Matching
57mJava Value Types: what is it and why are they important
55mThe Art of Java Language Pattern Matching