#VoxxedDaysCERN26

Java programmers ❤️ streams! They brought us at least some of the joys of functional programming: code that’s clear and elegant, and even, sometimes, performant. But how do they work? Many people don’t know, and that puts them in danger of writing code that seriously underperforms. In this talk we’ll look at the principles and some details of the implementation, and we’ll work out some useful principles about how – and even whether – to use streams in practice. The talk will be up to date with the new feature of Gatherers, revealing the mystery of the greedy flag.

You’ll come away from this talk with a new understanding of how streams work, the assumptions that underlie their design, and how you too can write stream code that both looks good and performs well.

Devoxx

A conference talk transcript by Java Champions Cay Horstmann and Maurice Naftalin exploring Java Streams internals and performance. They walk through how stream pipelines are constructed lazily, how terminal operations trigger computation via spliterators, and how the JIT optimizes certain pipeline shapes. Key performance topics include: the overhead of short-circuiting operations like limit, memory costs of distinct and sorted, autoboxing pitfalls with primitive streams, and when parallel streams help vs. hurt (compute-intensive workloads with large data sets are ideal; pipelines with filter+limit or iterate-based sources fail to parallelize effectively). The talk also introduces the Java 23 Gatherers API for custom intermediate operations, with examples like windowed grouping and index-pairing, and explains why some gatherers cannot be parallelized.

Streams – Below the Surface by Cay Horstmann and Maurice Naftalin