Concurrency

Starting a platform thread

Thread thread = new Thread(() -> System.out.println("working"));
thread.start();
thread.join();

Calling run() directly does not start a new thread.

Executors

try (ExecutorService executor = Executors.newFixedThreadPool(4)) {
    Future<Integer> result = executor.submit(() -> 40 + 2);
    System.out.println(result.get());
}

Executors separate task submission from thread creation and scheduling.

Shared state

synchronized void increment() {
    count++;
}

count++ is a read-modify-write sequence, not an atomic operation. Alternatives include locks and atomic variables.

AtomicInteger count = new AtomicInteger();
count.incrementAndGet();

CompletableFuture

CompletableFuture<String> result = CompletableFuture
        .supplyAsync(() -> "java")
        .thenApply(String::toUpperCase)
        .exceptionally(exception -> "UNKNOWN");

Use thenCompose for dependent asynchronous stages and thenCombine for independent results.

Virtual threads

Thread thread = Thread.startVirtualThread(() -> blockingOperation());
thread.join();

Virtual threads are suitable for high-concurrency workloads that spend much of their time blocked on I/O. They do not make CPU-intensive work execute faster.

Preview and internal examples

The repository also contains structured-concurrency, scoped-value, and continuation experiments. Some require --enable-preview; continuation examples use internal jdk.internal.vm APIs and should not be treated as supported application APIs.

Repository examples

• src/main/java/nitin/multithreading
• src/main/java/nitin/multithreading/cVirtualThreads
• src/test/java/nitin/multithreading
• src/main/java/nitin/multithreading/blog/BACKUP