We Keep Coding

Thread execution of value emitting code and value receiving code in RxJava

I have following code:
private static void log(Object msg) {
System.out.println(
Thread.currentThread().getName() +
": " + msg);
}
Observable<Integer> naturalNumbers = Observable.create(emitter -> {
log("Invoked"); // on main thread
Runnable r = () -> {
log("Invoked on another thread");
int i = 0;
while(!emitter.isDisposed()) {
log("Emitting "+ i);
emitter.onNext(i);
i += 1;
}
};
new Thread(r).start();
});
Disposable disposable = naturalNumbers.subscribe(i -> log("Received "+i));
So here we have 2 important lambda expressions. First is the one we pass to Observable.create, second is the callback one we pass to Observable.subscribe(). In first lambda, we create a new thread and then emit values on that thread. In second lambda, we have the code to receive those values emitted in first lambda code. I observe that both code are executed on same thread.
Thread-0: Invoked on another thread
Thread-0: Emitting 0
Thread-0: Received 0
Thread-0: Emitting 1
Thread-0: Received 1
Thread-0: Emitting 2
Thread-0: Received 2
Why is it so? Does RxJava by default run code emitting values(observable) and the code receiving values(observer) on same thread?

Let's see, what happens, if you use a Thread to execute a runnable:
Test
#Test
void threadTest() throws Exception {
log("main");
CountDownLatch countDownLatch = new CountDownLatch(1);
new Thread(
() -> {
log("thread");
countDownLatch.countDown();
})
.start();
countDownLatch.await();
}
Output
main: main
Thread-0: thread
It seems, that the main entry point is called from main thread and the newly created Thread is called Thread-0.
Why is it so? Does RxJava by default run code emitting values(observable) and the code receiving values(observer) on same thread?
By default RxJava is single-threaded. Therefore the the producer, if not definied differently by observeOn, subscribeOn or different threading layout, will emit values on the consumer (subsriber)-thread. This is because RxJava runs everything on the subscribing stack by default.
Example 2
#Test
void fdskfkjsj() throws Exception {
log("main");
Observable<Integer> naturalNumbers =
Observable.create(
emitter -> {
log("Invoked"); // on main thread
Runnable r =
() -> {
log("Invoked on another thread");
int i = 0;
while (!emitter.isDisposed()) {
log("Emitting " + i);
emitter.onNext(i);
i += 1;
}
};
new Thread(r).start();
});
Disposable disposable = naturalNumbers.subscribe(i -> log("Received " + i));
Thread.sleep(100);
}
Output2
main: main
main: Invoked
Thread-0: Invoked on another thread
Thread-0: Emitting 0
Thread-0: Received 0
Thread-0: Emitting 1
In your example it is apparent, that the main method is called from the main thread. Furthermore the subscribeActual call is also run on the calling-thread (main). But the Observable#create lambda calls onNext from the newly created thread Thread-0. The value is pushed to the subscriber from the calling thread. In this case, the calling thread is Thread-0, because it calls onNext on the downstream subscriber.
How to separate producer from consumer?
Use observeOn/ subscribeOn operators in order to handle concurrency in RxJava.
Should I use low-level Thread constructs ẁith RxJava?
No you should not use new Thread in order to seperate the producer from the consumer. It is quite easy to break the contract, that onNext can not be called concurrently (interleaving) and therefore breaking the contract. This is why RxJava provides a construct called Scheduler with Workers in order to mitigate such mistakes.
Note:
I think this article describes it quite well: http://introtorx.com/Content/v1.0.10621.0/15_SchedulingAndThreading.html . Please note this is Rx.NET, but the principle is quite the same. If you want to read about concurrency with RxJava you could also look into Davids Blog (https://akarnokd.blogspot.com/2015/05/schedulers-part-1.html) or read this Book (Reactive Programming with RxJava https://www.oreilly.com/library/view/reactive-programming-with/9781491931646/)

ExecutorService.submit() vs ExecutorSerivce.invokeXyz()

ExecutorService contains following methods:
invokeAll(Collection<? extends Callable<T>> tasks)
invokeAny(Collection<? extends Callable<T>> tasks)
submit(Callable<T> task)
I am confused about the use of terms submit vs invoke. Does it mean that invokeXyz() methods invoke those tasks immediately as soon as possible by underlying thread pool and submit() does some kind of scheduling of tasks submitted.
This answer says "if we want to wait for completion of all tasks, which have been submitted to ExecutorService". What does "wait for" here refers to?

Both invoke..() and submit() will execute their tasks immediately (assuming threads are available to run the tasks). The difference is that invoke...() will wait for the tasks running in separate threads to finish before returning a result, whereas submit() will return immediately, meaning the task it executed is still running in another thread.
In other words, the Future objects returned from invokeAll() are guaranteed to be in a state where Future.isDone() == true. The Future object returned from submit() can be in a state where Future.isDone() == false.
We can easily demonstrate the timing difference.
public static void main(String... args) throws InterruptedException {
Callable<String> c1 = () -> { System.out.println("Hello "); return "Hello "; };
Callable<String> c2 = () -> { System.out.println("World!"); return "World!"; };
List<Callable<String>> callables = List.of(c1, c2);
ExecutorService executor = Executors.newSingleThreadExecutor();
System.out.println("Begin invokeAll...");
List<Future<String>> allFutures = executor.invokeAll(callables);
System.out.println("End invokeAll.\n");
System.out.println("Begin submit...");
List<Future<String>> submittedFutures = callables.stream().map(executor::submit).collect(toList());
System.out.println("End submit.");
}
And the result is that the callables print their Hello World message before the invokeAll() method completes; but the callables print Hello World after the submit() method completes.
/*
Begin invokeAll...
Hello
World!
End invokeAll.
Begin submit...
End submit.
Hello
World!
*/
You can play around with this code in an IDE by adding some sleep() time in c1 or c2 and watching as the terminal prints out. This should convince you that invoke...() does indeed wait for something to happen, but submit() does not.

Java 8 Concurrency Simplest Canonical Form for Basic Task

I have two questions:
1. What is the simplest canonical form for running a Callable as a task in Java 8, capturing and processing the result?
2. In the example below, what is the best/simplest/clearest way to hold the main process open until all the tasks have completed?
Here's the example I have so far -- is this the best approach in Java 8 or is there something more basic?
import java.util.*;
import java.util.concurrent.*;
import java.util.function.*;
public class SimpleTask implements Supplier<String> {
private SplittableRandom rand = new SplittableRandom();
final int id;
SimpleTask(int id) { this.id = id; }
#Override
public String get() {
try {
TimeUnit.MILLISECONDS.sleep(rand.nextInt(50, 300));
} catch(InterruptedException e) {
System.err.println("Interrupted");
}
return "Completed " + id + " on " +
Thread.currentThread().getName();
}
public static void main(String[] args) throws Exception {
for(int i = 0; i < 10; i++)
CompletableFuture.supplyAsync(new SimpleTask(i))
.thenAccept(System.out::println);
System.in.read(); // Or else program ends too soon
}
}
Is there a simpler and clearer Java-8 way to do this? And how do I eliminate the System.in.read() in favor of a better approach?

The canonical way to wait for the completion of multiple CompletableFuture instance is to create a new one depending on all of them via CompletableFuture.allOf. You can use this new future to wait for its completion or schedule new follow-up actions just like with any other CompletableFuture:
CompletableFuture.allOf(
IntStream.range(0,10).mapToObj(SimpleTask::new)
.map(s -> CompletableFuture.supplyAsync(s).thenAccept(System.out::println))
.toArray(CompletableFuture<?>[]::new)
).join();
Of course, it always gets simpler if you forego assigning a unique id to each task. Since your first question was about Callable, I’ll demonstrate how you can easily submit multiple similar tasks as Callables via an ExecutorService:
ExecutorService pool = Executors.newCachedThreadPool();
pool.invokeAll(Collections.nCopies(10, () -> {
LockSupport.parkNanos(TimeUnit.MILLISECONDS.toNanos(
ThreadLocalRandom.current().nextInt(50, 300)));
final String s = "Completed on "+Thread.currentThread().getName();
System.out.println(s);
return s;
}));
pool.shutdown();
The executor service returned by Executors.newCachedThreadPool() is unshared and won’t stay alive, even if you forget to invoke shutDown(), but it can take up to one minute before all threads are terminated then.
Since your first question literally was: “What is the simplest canonical form for running a Callable as a task in Java 8, capturing and processing the result?”, the answer might be that the simplest form still is invoking it’s call() method directly, e.g.
Callable<String> c = () -> {
LockSupport.parkNanos(TimeUnit.MILLISECONDS.toNanos(
ThreadLocalRandom.current().nextInt(50, 300)));
return "Completed on "+Thread.currentThread().getName();
};
String result = c.call();
System.out.println(result);
There’s no simpler way…

Consider collecting the futures into a list. Then you can use join() on each future to await their completion in the current thread:
List<CompletableFuture<Void>> futures = IntStream.range(0,10)
.mapToObj(id -> supplyAsync(new SimpleTask(id)).thenAccept(System.out::println))
.collect(toList());
futures.forEach(CompletableFuture::join);

wait until all threads finish their work in java

I'm writing an application that has 5 threads that get some information from web simultaneously and fill 5 different fields in a buffer class.
I need to validate buffer data and store it in a database when all threads finished their job.
How can I do this (get alerted when all threads finished their work) ?

The approach I take is to use an ExecutorService to manage pools of threads.
ExecutorService es = Executors.newCachedThreadPool();
for(int i=0;i<5;i++)
es.execute(new Runnable() { /* your task */ });
es.shutdown();
boolean finished = es.awaitTermination(1, TimeUnit.MINUTES);
// all tasks have finished or the time has been reached.

You can join to the threads. The join blocks until the thread completes.
for (Thread thread : threads) {
thread.join();
}
Note that join throws an InterruptedException. You'll have to decide what to do if that happens (e.g. try to cancel the other threads to prevent unnecessary work being done).

Have a look at various solutions.
join() API has been introduced in early versions of Java. Some good alternatives are available with this concurrent package since the JDK 1.5 release.
ExecutorService#invokeAll()
Executes the given tasks, returning a list of Futures holding their status and results when everything is completed.
Refer to this related SE question for code example:
How to use invokeAll() to let all thread pool do their task?
CountDownLatch
A synchronization aid that allows one or more threads to wait until a set of operations being performed in other threads completes.
A CountDownLatch is initialized with a given count. The await methods block until the current count reaches zero due to invocations of the countDown() method, after which all waiting threads are released and any subsequent invocations of await return immediately. This is a one-shot phenomenon -- the count cannot be reset. If you need a version that resets the count, consider using a CyclicBarrier.
Refer to this question for usage of CountDownLatch
How to wait for a thread that spawns it's own thread?
ForkJoinPool or newWorkStealingPool() in Executors
Iterate through all Future objects created after submitting to ExecutorService

Wait/block the Thread Main until some other threads complete their work.
As #Ravindra babu said it can be achieved in various ways, but showing with examples.
java.lang.Thread.join() Since:1.0
public static void joiningThreads() throws InterruptedException {
Thread t1 = new Thread( new LatchTask(1, null), "T1" );
Thread t2 = new Thread( new LatchTask(7, null), "T2" );
Thread t3 = new Thread( new LatchTask(5, null), "T3" );
Thread t4 = new Thread( new LatchTask(2, null), "T4" );
// Start all the threads
t1.start();
t2.start();
t3.start();
t4.start();
// Wait till all threads completes
t1.join();
t2.join();
t3.join();
t4.join();
}
java.util.concurrent.CountDownLatch Since:1.5
.countDown() « Decrements the count of the latch group.
.await() « The await methods block until the current count reaches zero.
If you created latchGroupCount = 4 then countDown() should be called 4 times to make count 0. So, that await() will release the blocking threads.
public static void latchThreads() throws InterruptedException {
int latchGroupCount = 4;
CountDownLatch latch = new CountDownLatch(latchGroupCount);
Thread t1 = new Thread( new LatchTask(1, latch), "T1" );
Thread t2 = new Thread( new LatchTask(7, latch), "T2" );
Thread t3 = new Thread( new LatchTask(5, latch), "T3" );
Thread t4 = new Thread( new LatchTask(2, latch), "T4" );
t1.start();
t2.start();
t3.start();
t4.start();
//latch.countDown();
latch.await(); // block until latchGroupCount is 0.
}
Example code of Threaded class LatchTask. To test the approach use joiningThreads();
and latchThreads(); from main method.
class LatchTask extends Thread {
CountDownLatch latch;
int iterations = 10;
public LatchTask(int iterations, CountDownLatch latch) {
this.iterations = iterations;
this.latch = latch;
}
#Override
public void run() {
String threadName = Thread.currentThread().getName();
System.out.println(threadName + " : Started Task...");
for (int i = 0; i < iterations; i++) {
System.out.println(threadName + " : " + i);
MainThread_Wait_TillWorkerThreadsComplete.sleep(1);
}
System.out.println(threadName + " : Completed Task");
// countDown() « Decrements the count of the latch group.
if(latch != null)
latch.countDown();
}
}
CyclicBarriers A synchronization aid that allows a set of threads to all wait for each other to reach a common barrier point.CyclicBarriers are useful in programs involving a fixed sized party of threads that must occasionally wait for each other. The barrier is called cyclic because it can be re-used after the waiting threads are released.
CyclicBarrier barrier = new CyclicBarrier(3);
barrier.await();
For example refer this Concurrent_ParallelNotifyies class.
Executer framework: we can use ExecutorService to create a thread pool, and tracks the progress of the asynchronous tasks with Future.
submit(Runnable), submit(Callable) which return Future Object. By using future.get() function we can block the main thread till the working threads completes its work.
invokeAll(...) - returns a list of Future objects via which you can obtain the results of the executions of each Callable.
Find example of using Interfaces Runnable, Callable with Executor framework.
#See also
Find out thread is still alive?

Apart from Thread.join() suggested by others, java 5 introduced the executor framework. There you don't work with Thread objects. Instead, you submit your Callable or Runnable objects to an executor. There's a special executor that is meant to execute multiple tasks and return their results out of order. That's the ExecutorCompletionService:
ExecutorCompletionService executor;
for (..) {
executor.submit(Executors.callable(yourRunnable));
}
Then you can repeatedly call take() until there are no more Future<?> objects to return, which means all of them are completed.
Another thing that may be relevant, depending on your scenario is CyclicBarrier.
A synchronization aid that allows a set of threads to all wait for each other to reach a common barrier point. CyclicBarriers are useful in programs involving a fixed sized party of threads that must occasionally wait for each other. The barrier is called cyclic because it can be re-used after the waiting threads are released.

Another possibility is the CountDownLatch object, which is useful for simple situations : since you know in advance the number of threads, you initialize it with the relevant count, and pass the reference of the object to each thread.
Upon completion of its task, each thread calls CountDownLatch.countDown() which decrements the internal counter. The main thread, after starting all others, should do the CountDownLatch.await() blocking call. It will be released as soon as the internal counter has reached 0.
Pay attention that with this object, an InterruptedException can be thrown as well.

You do
for (Thread t : new Thread[] { th1, th2, th3, th4, th5 })
t.join()
After this for loop, you can be sure all threads have finished their jobs.

Store the Thread-objects into some collection (like a List or a Set), then loop through the collection once the threads are started and call join() on the Threads.

You can use Threadf#join method for this purpose.

Although not relevant to OP's problem, if you are interested in synchronization (more precisely, a rendez-vous) with exactly one thread, you may use an Exchanger
In my case, I needed to pause the parent thread until the child thread did something, e.g. completed its initialization. A CountDownLatch also works well.

I created a small helper method to wait for a few Threads to finish:
public static void waitForThreadsToFinish(Thread... threads) {
try {
for (Thread thread : threads) {
thread.join();
}
}
catch (InterruptedException e) {
e.printStackTrace();
}
}

An executor service can be used to manage multiple threads including status and completion. See http://programmingexamples.wikidot.com/executorservice

try this, will work.
Thread[] threads = new Thread[10];
List<Thread> allThreads = new ArrayList<Thread>();
for(Thread thread : threads){
if(null != thread){
if(thread.isAlive()){
allThreads.add(thread);
}
}
}
while(!allThreads.isEmpty()){
Iterator<Thread> ite = allThreads.iterator();
while(ite.hasNext()){
Thread thread = ite.next();
if(!thread.isAlive()){
ite.remove();
}
}
}

I had a similar problem and ended up using Java 8 parallelStream.
requestList.parallelStream().forEach(req -> makeRequest(req));
It's super simple and readable.
Behind the scenes it is using default JVM’s fork join pool which means that it will wait for all the threads to finish before continuing. For my case it was a neat solution, because it was the only parallelStream in my application. If you have more than one parallelStream running simultaneously, please read the link below.
More information about parallel streams here.

The existing answers said could join() each thread.
But there are several ways to get the thread array / list:
Add the Thread into a list on creation.
Use ThreadGroup to manage the threads.
Following code will use the ThreadGruop approach. It create a group first, then when create each thread specify the group in constructor, later could get the thread array via ThreadGroup.enumerate()
Code
SyncBlockLearn.java
import org.testng.Assert;
import org.testng.annotations.Test;
/**
* synchronized block - learn,
*
* #author eric
* #date Apr 20, 2015 1:37:11 PM
*/
public class SyncBlockLearn {
private static final int TD_COUNT = 5; // thread count
private static final int ROUND_PER_THREAD = 100; // round for each thread,
private static final long INC_DELAY = 10; // delay of each increase,
// sync block test,
#Test
public void syncBlockTest() throws InterruptedException {
Counter ct = new Counter();
ThreadGroup tg = new ThreadGroup("runner");
for (int i = 0; i < TD_COUNT; i++) {
new Thread(tg, ct, "t-" + i).start();
}
Thread[] tArr = new Thread[TD_COUNT];
tg.enumerate(tArr); // get threads,
// wait all runner to finish,
for (Thread t : tArr) {
t.join();
}
System.out.printf("\nfinal count: %d\n", ct.getCount());
Assert.assertEquals(ct.getCount(), TD_COUNT * ROUND_PER_THREAD);
}
static class Counter implements Runnable {
private final Object lkOn = new Object(); // the object to lock on,
private int count = 0;
#Override
public void run() {
System.out.printf("[%s] begin\n", Thread.currentThread().getName());
for (int i = 0; i < ROUND_PER_THREAD; i++) {
synchronized (lkOn) {
System.out.printf("[%s] [%d] inc to: %d\n", Thread.currentThread().getName(), i, ++count);
}
try {
Thread.sleep(INC_DELAY); // wait a while,
} catch (InterruptedException e) {
e.printStackTrace();
}
}
System.out.printf("[%s] end\n", Thread.currentThread().getName());
}
public int getCount() {
return count;
}
}
}
The main thread will wait for all threads in the group to finish.

I had similar situation , where i had to wait till all child threads complete its execution then only i could get the status result for each of them .. hence i needed to wait till all child thread completed.
below is my code where i did multi-threading using
public static void main(String[] args) {
List<RunnerPojo> testList = ExcelObject.getTestStepsList();//.parallelStream().collect(Collectors.toList());
int threadCount = ConfigFileReader.getInstance().readConfig().getParallelThreadCount();
System.out.println("Thread count is : ========= " + threadCount); // 5
ExecutorService threadExecutor = new DriverScript().threadExecutor(testList, threadCount);
boolean isProcessCompleted = waitUntilCondition(() -> threadExecutor.isTerminated()); // Here i used waitUntil condition
if (isProcessCompleted) {
testList.forEach(x -> {
System.out.println("Test Name: " + x.getTestCaseId());
System.out.println("Test Status : " + x.getStatus());
System.out.println("======= Test Steps ===== ");
x.getTestStepsList().forEach(y -> {
System.out.println("Step Name: " + y.getDescription());
System.out.println("Test caseId : " + y.getTestCaseId());
System.out.println("Step Status: " + y.getResult());
System.out.println("\n ============ ==========");
});
});
}
Below method is for distribution of list with parallel proccessing
// This method will split my list and run in a parallel process with mutliple threads
private ExecutorService threadExecutor(List<RunnerPojo> testList, int threadSize) {
ExecutorService exec = Executors.newFixedThreadPool(threadSize);
testList.forEach(tests -> {
exec.submit(() -> {
driverScript(tests);
});
});
exec.shutdown();
return exec;
}
This is my wait until method: here you can wait till your condition satisfies within do while loop . in my case i waited for some max timeout .
this will keep checking until your threadExecutor.isTerminated() is true with polling period of 5 sec.
static boolean waitUntilCondition(Supplier<Boolean> function) {
Double timer = 0.0;
Double maxTimeOut = 20.0;
boolean isFound;
do {
isFound = function.get();
if (isFound) {
break;
} else {
try {
Thread.sleep(5000); // Sleeping for 5 sec (main thread will sleep for 5 sec)
} catch (InterruptedException e) {
e.printStackTrace();
}
timer++;
System.out.println("Waiting for condition to be true .. waited .." + timer * 5 + " sec.");
}
} while (timer < maxTimeOut + 1.0);
return isFound;
}

Use this in your main thread: while(!executor.isTerminated());
Put this line of code after starting all the threads from executor service. This will only start the main thread after all the threads started by executors are finished. Make sure to call executor.shutdown(); before the above loop.

Java ExecutorService: awaitTermination of all recursively created tasks

I use an ExecutorService to execute a task. This task can recursively create other tasks which are submitted to the same ExecutorService and those child tasks can do that, too.
I now have the problem that I want to wait until all the tasks are done (that is, all tasks are finished and they did not submit new ones) before I continue.
I cannot call ExecutorService.shutdown() in the main thread because this prevents new tasks from being accepted by the ExecutorService.
And Calling ExecutorService.awaitTermination() seems to do nothing if shutdown hasn't been called.
So I am kinda stuck here. It can't be that hard for the ExecutorService to see that all workers are idle, can it? The only inelegant solution I could come up with is to directly use a ThreadPoolExecutor and query its getPoolSize() every once in a while. Is there really no better way do do that?

This really is an ideal candidate for a Phaser. Java 7 is coming out with this new class. Its a flexible CountdonwLatch/CyclicBarrier. You can get a stable version at JSR 166 Interest Site.
The way it is a more flexible CountdownLatch/CyclicBarrier is because it is able to not only support an unknown number of parties (threads) but its also reusable (thats where the phase part comes in)
For each task you submit you would register, when that task is completed you arrive. This can be done recursively.
Phaser phaser = new Phaser();
ExecutorService e = //
Runnable recursiveRunnable = new Runnable(){
public void run(){
//do work recursively if you have to
if(shouldBeRecursive){
phaser.register();
e.submit(recursiveRunnable);
}
phaser.arrive();
}
}
public void doWork(){
int phase = phaser.getPhase();
phaser.register();
e.submit(recursiveRunnable);
phaser.awaitAdvance(phase);
}
Edit: Thanks #depthofreality for pointing out the race condition in my previous example. I am updating it so that executing thread only awaits advance of the current phase as it blocks for the recursive function to complete.
The phase number won't trip until the number of arrives == registers. Since prior to each recursive call invokes register a phase increment will happen when all invocations are complete.

If number of tasks in the tree of recursive tasks is initially unknown, perhaps the easiest way would be to implement your own synchronization primitive, some kind of "inverse semaphore", and share it among your tasks. Before submitting each task you increment a value, when task is completed, it decrements that value, and you wait until the value is 0.
Implementing it as a separate primitive explicitly called from tasks decouples this logic from the thread pool implementation and allows you to submit several independent trees of recursive tasks into the same pool.
Something like this:
public class InverseSemaphore {
private int value = 0;
private Object lock = new Object();
public void beforeSubmit() {
synchronized(lock) {
value++;
}
}
public void taskCompleted() {
synchronized(lock) {
value--;
if (value == 0) lock.notifyAll();
}
}
public void awaitCompletion() throws InterruptedException {
synchronized(lock) {
while (value > 0) lock.wait();
}
}
}
Note that taskCompleted() should be called inside a finally block, to make it immune to possible exceptions.
Also note that beforeSubmit() should be called by the submitting thread before the task is submitted, not by the task itself, to avoid possible "false completion" when old tasks are completed and new ones not started yet.
EDIT: Important problem with usage pattern fixed.

Wow, you guys are quick:)
Thank you for all the suggestions. Futures don't easily integrate with my model because I don't know how many runnables are scheduled beforehand. So if I keep a parent task alive just to wait for it's recursive child tasks to finish I have a lot of garbage laying around.
I solved my problem using the AtomicInteger suggestion. Essentially, I subclassed ThreadPoolExecutor and increment the counter on calls to execute() and decrement on calls to afterExecute(). When the counter gets 0 I call shutdown(). This seems to work for my problems, not sure if that's a generally good way to do that. Especially, I assume that you only use execute() to add Runnables.
As a side node: I first tried to check in afterExecute() the number of Runnables in the queue and the number of workers that are active and shutdown when those are 0; but that didn't work because not all Runnables showed up in the queue and the getActiveCount() didn't do what I expected either.
Anyhow, here's my solution: (if anybody finds serious problems with this, please let me know:)
public class MyThreadPoolExecutor extends ThreadPoolExecutor {
private final AtomicInteger executing = new AtomicInteger(0);
public MyThreadPoolExecutor(int coorPoolSize, int maxPoolSize, long keepAliveTime,
TimeUnit seconds, BlockingQueue<Runnable> queue) {
super(coorPoolSize, maxPoolSize, keepAliveTime, seconds, queue);
}
#Override
public void execute(Runnable command) {
//intercepting beforeExecute is too late!
//execute() is called in the parent thread before it terminates
executing.incrementAndGet();
super.execute(command);
}
#Override
protected void afterExecute(Runnable r, Throwable t) {
super.afterExecute(r, t);
int count = executing.decrementAndGet();
if(count == 0) {
this.shutdown();
}
}
}

You could create your own thread pool which extends ThreadPoolExecutor. You want to know when a task has been submitted and when it completes.
public class MyThreadPoolExecutor extends ThreadPoolExecutor {
private int counter = 0;
public MyThreadPoolExecutor() {
super(1, 1, 0, TimeUnit.SECONDS, new LinkedBlockingQueue<Runnable>());
}
#Override
public synchronized void execute(Runnable command) {
counter++;
super.execute(command);
}
#Override
protected synchronized void afterExecute(Runnable r, Throwable t) {
super.afterExecute(r, t);
counter--;
notifyAll();
}
public synchronized void waitForExecuted() throws InterruptedException {
while (counter == 0)
wait();
}
}

Use a Future for your tasks (instead of submitting Runnable's), a callback updates it's state when it's completed, so you can use Future.isDone to track the sate of all your tasks.

(mea culpa: its a 'bit' past my bedtime ;) but here's a first attempt at a dynamic latch):
package oss.alphazero.sto4958330;
import java.util.concurrent.CountDownLatch;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.locks.AbstractQueuedSynchronizer;
public class DynamicCountDownLatch {
#SuppressWarnings("serial")
private static final class Sync extends AbstractQueuedSynchronizer {
private final CountDownLatch toplatch;
public Sync() {
setState(0);
this.toplatch = new CountDownLatch(1);
}
#Override
protected int tryAcquireShared(int acquires){
try {
toplatch.await();
}
catch (InterruptedException e) {
throw new RuntimeException("Interrupted", e);
}
return getState() == 0 ? 1 : -1;
}
public boolean tryReleaseShared(int releases) {
for (;;) {
int c = getState();
if (c == 0)
return false;
int nextc = c-1;
if (compareAndSetState(c, nextc))
return nextc == 0;
}
}
public boolean tryExtendState(int acquires) {
for (;;) {
int s = getState();
int exts = s+1;
if (compareAndSetState(s, exts)) {
toplatch.countDown();
return exts > 0;
}
}
}
}
private final Sync sync;
public DynamicCountDownLatch(){
this.sync = new Sync();
}
public void await()
throws InterruptedException
{
sync.acquireSharedInterruptibly(1);
}
public boolean await(long timeout, TimeUnit unit)
throws InterruptedException
{
return sync.tryAcquireSharedNanos(1, unit.toNanos(timeout));
}
public void countDown() {
sync.releaseShared(1);
}
public void join() {
sync.tryExtendState(1);
}
}
This latch introduces a new method join() to the existing (cloned) CountDownLatch API, which is used by tasks to signal their entry into the larger task group.
The latch is pass around from parent Task to child Task. Each task would, per Suraj's pattern, first 'join()' the latch, do its task(), and then countDown().
To address situations where the main thread launches the task group and then immediately awaits() -- before any of the task threads have had a chance to even join() -- the topLatch is used int inner Sync class. This is a latch that will get counted down on each join(); only the first countdown is of course significant, as all subsequent ones are nops.
The initial implementation above does introduce a semantic wrinkle of sorts since the tryAcquiredShared(int) is not supposed to be throwing an InterruptedException but then we do need to deal with the interrupt on the wait on the topLatch.
Is this an improvement over OP's own solution using Atomic counters? I would say probably not IFF he is insistent upon using Executors, but it is, I believe, an equally valid alternative approach using the AQS in that case, and, is usable with generic threads as well.
Crit away fellow hackers.

If you want to use JSR166y classes - e.g. Phaser or Fork/Join - either of which might work for you, you can always download the Java 6 backport of them from: http://gee.cs.oswego.edu/dl/concurrency-interest/ and use that as a basis rather than writing a completely homebrew solution. Then when 7 comes out you can just drop the dependency on the backport and change a few package names.
(Full disclosure: We've been using the LinkedTransferQueue in prod for a while now. No issues)

I must say, that solutions described above of problem with recursive calling task and wait for end suborder tasks doesn't satisfy me. There is my solution inspired by original documentation from Oracle there: CountDownLatch and example there: Human resources CountDownLatch.
The first common thread in process in instance of class HRManagerCompact has waiting latch for two daughter's threads, wich has waiting latches for their subsequent 2 daughter's threads... etc.
Of course, latch can be set on the different value than 2 (in constructor of CountDownLatch), as well as the number of runnable objects can be established in iteration i.e. ArrayList, but it must correspond (number of count downs must be equal the parameter in CountDownLatch constructor).
Be careful, the number of latches increases exponentially according restriction condition:
'level.get() < 2', as well as the number of objects. 1, 2, 4, 8, 16... and latches 0, 1, 2, 4... As you can see, for four levels (level.get() < 4) there will be 15 waiting threads and 7 latches in the time, when peak 16 threads are running.
package processes.countdownlatch.hr;
import java.util.concurrent.CountDownLatch;
import java.util.concurrent.atomic.AtomicInteger;
import java.util.concurrent.atomic.AtomicLong;
/** Recursively latching running classes to wait for the peak threads
*
* #author hariprasad
*/
public class HRManagerCompact extends Thread {
final int N = 2; // number of daughter's tasks for latch
CountDownLatch countDownLatch;
CountDownLatch originCountDownLatch;
AtomicInteger level = new AtomicInteger(0);
AtomicLong order = new AtomicLong(0); // id latched thread waiting for
HRManagerCompact techLead1 = null;
HRManagerCompact techLead2 = null;
HRManagerCompact techLead3 = null;
// constructor
public HRManagerCompact(CountDownLatch countDownLatch, String name,
AtomicInteger level, AtomicLong order){
super(name);
this.originCountDownLatch=countDownLatch;
this.level = level;
this.order = order;
}
private void doIt() {
countDownLatch = new CountDownLatch(N);
AtomicInteger leveli = new AtomicInteger(level.get() + 1);
AtomicLong orderi = new AtomicLong(Thread.currentThread().getId());
techLead1 = new HRManagerCompact(countDownLatch, "first", leveli, orderi);
techLead2 = new HRManagerCompact(countDownLatch, "second", leveli, orderi);
//techLead3 = new HRManagerCompact(countDownLatch, "third", leveli);
techLead1.start();
techLead2.start();
//techLead3.start();
try {
synchronized (Thread.currentThread()) { // to prevent print and latch in the same thread
System.out.println("*** HR Manager waiting for recruitment to complete... " + level + ", " + order + ", " + orderi);
countDownLatch.await(); // wait actual thread
}
System.out.println("*** Distribute Offer Letter, it means finished. " + level + ", " + order + ", " + orderi);
} catch (InterruptedException e) {
e.printStackTrace();
}
}
#Override
public void run() {
try {
System.out.println(Thread.currentThread().getName() + ": working... " + level + ", " + order + ", " + Thread.currentThread().getId());
Thread.sleep(10*level.intValue());
if (level.get() < 2) doIt();
Thread.yield();
}
catch (Exception e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
/*catch (InterruptedException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}*/
// TODO Auto-generated method stub
System.out.println("--- " +Thread.currentThread().getName() + ": recruted " + level + ", " + order + ", " + Thread.currentThread().getId());
originCountDownLatch.countDown(); // count down
}
public static void main(String args[]){
AtomicInteger levelzero = new AtomicInteger(0);
HRManagerCompact hr = new HRManagerCompact(null, "zero", levelzero, new AtomicLong(levelzero.longValue()));
hr.doIt();
}
}
Possible commented output (with some probability):
first: working... 1, 1, 10 // thread 1, first daughter's task (10)
second: working... 1, 1, 11 // thread 1, second daughter's task (11)
first: working... 2, 10, 12 // thread 10, first daughter's task (12)
first: working... 2, 11, 14 // thread 11, first daughter's task (14)
second: working... 2, 11, 15 // thread 11, second daughter's task (15)
second: working... 2, 10, 13 // thread 10, second daughter's task (13)
--- first: recruted 2, 10, 12 // finished 12
--- first: recruted 2, 11, 14 // finished 14
--- second: recruted 2, 10, 13 // finished 13 (now can be opened latch 10)
--- second: recruted 2, 11, 15 // finished 15 (now can be opened latch 11)
*** HR Manager waiting for recruitment to complete... 0, 0, 1
*** HR Manager waiting for recruitment to complete... 1, 1, 10
*** Distribute Offer Letter, it means finished. 1, 1, 10 // latch on 10 opened
--- first: recruted 1, 1, 10 // finished 10
*** HR Manager waiting for recruitment to complete... 1, 1, 11
*** Distribute Offer Letter, it means finished. 1, 1, 11 // latch on 11 opened
--- second: recruted 1, 1, 11 // finished 11 (now can be opened latch 1)
*** Distribute Offer Letter, it means finished. 0, 0, 1 // latch on 1 opened

Use CountDownLatch.
Pass the CountDownLatch object to each of your tasks and code your tasks something like below.
public void doTask() {
// do your task
latch.countDown();
}
Whereas the thread which needs to wait should execute the following code:
public void doWait() {
latch.await();
}
But ofcourse, this assumes you already know the number of child tasks so that you could initialize the latch's count.

The only inelegant solution I could come up with is to directly use a ThreadPoolExecutor and query its getPoolSize() every once in a while. Is there really no better way do do that?
You have to use shutdown() ,awaitTermination()and shutdownNow() methods in a proper sequence.
shutdown(): Initiates an orderly shutdown in which previously submitted tasks are executed, but no new tasks will be accepted.
awaitTermination():Blocks until all tasks have completed execution after a shutdown request, or the timeout occurs, or the current thread is interrupted, whichever happens first.
shutdownNow(): Attempts to stop all actively executing tasks, halts the processing of waiting tasks, and returns a list of the tasks that were awaiting execution.
Recommended way from oracle documentation page of ExecutorService:
void shutdownAndAwaitTermination(ExecutorService pool) {
pool.shutdown(); // Disable new tasks from being submitted
try {
// Wait a while for existing tasks to terminate
if (!pool.awaitTermination(60, TimeUnit.SECONDS)) {
pool.shutdownNow(); // Cancel currently executing tasks
// Wait a while for tasks to respond to being cancelled
if (!pool.awaitTermination(60, TimeUnit.SECONDS))
System.err.println("Pool did not terminate");
}
} catch (InterruptedException ie) {
// (Re-)Cancel if current thread also interrupted
pool.shutdownNow();
// Preserve interrupt status
Thread.currentThread().interrupt();
}
You can replace if condition with while condition in case of long duration in completion of tasks as below:
Change
if (!pool.awaitTermination(60, TimeUnit.SECONDS))
To
while(!pool.awaitTermination(60, TimeUnit.SECONDS)) {
Thread.sleep(60000);
}
You can refer to other alternatives (except join(), which can be used with standalone thread ) in :
wait until all threads finish their work in java

You could use a runner that keeps track of running threads:
Runner runner = Runner.runner(numberOfThreads);
runner.runIn(2, SECONDS, callable);
runner.run(callable);
// blocks until all tasks are finished (or failed)
runner.waitTillDone();
// and reuse it
runner.runRunnableIn(500, MILLISECONDS, runnable);
runner.waitTillDone();
// and then just kill it
runner.shutdownAndAwaitTermination();
to use it you just add a dependency:
compile 'com.github.matejtymes:javafixes:1.3.0'