I have a set of jobs which I am submitting using executor framework and Future. Let's say that I have 100 futures. As of now, I am using Future.get and using the output for subsequent processing. However for further tuning, I want to change the flow as below:
iterate through the set of future tasks and start consuming the result as soon as a future task is complete. I am reading the API doc to understand what might be a good way to accomplish this but reaching out to see if there is a better way to accomplish what I am looking for.
Here is the sample code:
public class ImplCallable implements Callable<String> {
int timeOut;
ImplCallable(int timeOut) {
this.timeOut=timeOut;
}
public String call() throws Exception {
Thread.sleep(timeOut);
return Thread.currentThread().getName();
}
}
and the main class:
public class MainProg {
public static void main(String...args) throws Exception {
long startTimeInMillis = System.currentTimeMillis();
ImplCallable callable1 = new ImplCallable(1000);
ImplCallable callable2 = new ImplCallable(2000);
ExecutorService service = Executors.newFixedThreadPool(4);
Future<String> task1 = service.submit(callable1);
Future<String> task2 = service.submit(callable2);
List<Future<String>> futureList = new ArrayList();
futureList.add(task1);
futureList.add(task2);
String retVal;
for(Future<String> task:futureList) {
retVal = task.get();
//do something with the retVal
}
long endTimeInMillis = System.currentTimeMillis();
System.out.println("time taken by code - " + (endTimeInMillis-startTimeInMillis) + "-ms");
}
}
Basically I don't want to use Future.get() and wait for its completion. I want to know if either of the task is complete and use the result as soon as its done.
There are many ways do this so, without a concrete example, you won't get a concrete answer. Likely want to look at CompletableFuture which has many methods for defining follow-on work, combining work, splitting work etc.
Future<String> f = CompletableFuture.supplyAsync(() -> "INITIAL WORK")
.thenApply(String::toLowerCase) // Do some more work
.thenAccept(queue::add); // put results onto a queue something is reading from
f.join();
// Batch complete
I hope you are using Java 8 or later version.
Whenever you mention "as soon as a future task is complete", you want to use CompletableFuture and its .thenApply() method, as #drekbour suggests.
Then you have multiple threads running different tasks in non-determenistic sequence. But at the end you want to get all the results in the single (Main) thread. To achieve it, you can use CompletableFuture.allOf method, .join() it - and then iterate over all the (already completed) future results without waiting.
Related
Consider the following Java code (a simplified version of what I am working with - if there are errors it's because I haven't run it through a compiler):
CountdownLatch latch = new CountdownLatch(collection.size());
for(Whatever thing : collection){
provider.doWork(thing, result -> {
process(result);
latch.countDown();
};
}
try {
latch.await();
} catch (InterruptedException ignore) {}
doMoreWork();
So I run a number of asynchronous tasks, and wait for them all to be done before proceeding. Right now I'm accumulating the result of the asynchronous tasks in a list. This works and it's fine, but I'm looking at whether there's a cleaner implementation using Futures or something similar. The issue is the asynchronous call. A Callable is supposed to return the result of its work, but the result of this work won't be known until later. It's not worth rewriting doWork to be synchronous. Should I just leave this alone, or is there an option out there? Partly my interest is in better code but partly in just learning more about concurrency options. If it matters, this is in an Android app.
This is an approach using ExecutorService and Future, didn't tested on android but all available on level 1 api:
ExecutorService executor = Executors.newFixedThreadPool(10);
List<Future<String>> pending = new ArrayList<Future<String>>();
for(Whatever thing : collection) {
Future<String> future = executor.submit(new Callable<String>() {
public String call() throws Exception {
return doWork();
}
});
pending.add(future);
}
for (Future<String> result : pending) {
System.out.println("Your result ASAP:" + result.get());
}
executor.shutdown();
It will return as soon as the current item is done, in the order the were submitted.
void process(String question){
Callable<ResponseList> callable1 = () -> this.stsCompute question);
Future<ResponseList>task1 = executorService.submit(callable1);
Callable<ResponseList> callable2 = () -> this.dssmCompute(question);
Future<ResponseList>task2 = executorService.submit(callable2);
try{
ResponseList stsResponse = task1.get();
ResponseList dssmResponse = task2.get();
}catch (Exception e){
e.printStackTrace();
}
// Do I need to wait until the first 2 threads complete?
processResponse(stsResponse, dssmResponse);
}
In this "process" method, I have two additional threads 'callable1' & 'callable2' to concurrently execute. I want to make sure only when these two tasks complete, the method in the main thread 'processResponse()' can start to be executed.
In such a case, do I need to add any additional control to ensure the order of the execution, is it already good enough? If not, how to make that control happen?
You should use ExecutorService.invokeAll which will return List of Futures when complete. Besides I would use a shorter syntax, something like
List<Future> futures = executorService.invokeAll(Arrays.asList(()->dssmCompute(), ()->dssmCompute()));
With Java8+ i would suggest use Completable Futures. It supports exactly the use case you are trying to achieve.
Completable Futures: Java 8
Sample Algorithm looks like:
var cf = CompletableFuture.supplyAsync(() -> processQuestion()).runAsync(() -> processResponse)
Note: var is typeInference supports in java 10+
Also, there are plenty of Examples on Completable Futures
I am new to Multithreading and I am trying make my program faster using ExecutorService. Below is y implementation but, my program is still not working fast. Can you please look at my code and advise?
It basically reads the list of email addresses and stores in the ArrayList. I use the ExecutorService and loop through the ArrayList and call a Callable class does some processing and returns a Boolean.
ArrayList<String> emailAddressList = new ArrayList<>();
ExecutorService executor = Executors.newFixedThreadPool(7);
for (int i = 0; i < emailAddressList.size(); i++) {
Future<Boolean> resultFromThread = executor.submit(new Verify(emailAddressList.get(i)));
bufferedWriter.write(emailAddressList.get(i) + "|" + resultFromThread.get());
bufferedWriter.newLine();
}
executor.shutdown();
executor.awaitTermination(3, TimeUnit.SECONDS);
===========================================================================
public class Verify implements Callable<Boolean> {
private String email;
public Verify(String email) {
this.email = email;
}
#Override
public Boolean call() throws Exception {
Boolean result = false;
try {
result = Validator2.isAddressValid(email);
} catch (Exception e) {
}
return result;
}
}
In each iteration of the loop, two actions are performed:
A Callable is scheduled to run with the Executor
Immediately after that - yet before another task is scheduled - the code waits for the Executor to complete the Callable just submitted.
That way, all Callables are still executed in a serial fashion (we wait to complete one before we submit another), rather than executing them in parallel.
A simple solution might be to submit all callables for execution first. Then, in a separate loop, the wait for them to complete and to process the results. That way, the performance shall improve because of parallel processing of the Callables.
Example:
List<Future<Boolean>> futures ... ;
for (int i = 0; i < emailAddressList.size(); i++) {
futures.add(executor.submit(new Verify(emailAddressList.get(i))));
}
for (int i = 0; i < emailAddressList.size(); i++)
bufferedWriter.write(emailAddressList.get(i) + "|" + futures.get(i).get());
bufferedWriter.newLine();
}
Note that this code waits for the Callables to complete in the order they were submitted to the Executor. This may not necesarily be the case. If the order of the adresses in the resulting writer is not important, one may consider a completely asychronous processing. In Java 8, this can be achieved e.g using the CompleteableFuture API.
You have effectively made your code synchronous and single-threaded without offering any advantage to using a threaded executor. When calling resultFromThread.get(), it will block the main thread and will prevent the next loop iteration that submits the next Task to execute from running until the previous one completes. If you want the submitted Verify tasks to run concurrently, you should submit all of the tasks first in one loop appended to a List<Future<Boolean>>. Then, in another loop, you can iterate through each of those and then call .get(), such that the main thread will await completion of all executions, but won't stop the 7 other threads from executing concurrently.
Is there an ExecutorService that will allow me to submit tasks without beginning execution until I request it? I'm looking for something like ScheduledExecutorService, except that I want to trigger the execution manually without depending on a fixed time delay.
The reason I'm looking for this is because I want to create a set of tasks which may recursively use results of Futures generated from parallel tasks in the same set. So I would need to first submit all the tasks in order to get a set of Futures, and only then could I allow the tasks to begin executing.
It sounds like a job for CompletableFuture
fire execution of first portion of tasks as separe CompletableFutures
then using CompletableFuture.allOf(...furures) to create a barrier future that completes only when all are done
then using one of combinators like CompletableFuture.thenAccept to schedule next portion of tasks to perform on completion of barrier future
But more idiomatic way to use it would be to chain each next task based on future result of some previous one
CompletableFuture<FirstResult> firstTask = //....
CompletableFuture<SecondResult> secondTask = firstTask.thenApply(someTransformation);
CompletableFuture<Void> anotherTaks = firstTask.thenAccept(someConsumer);
CompletableFuture<ThirdResult> combined = firstTask.thenAcceptBoth(secondTask, someFunction);
Perhaps an alternate approach would be to simply use a FutureCallback or an AsyncFunction?
FutureCallback example:
final List<ListenableFuture<T>> futures = new ArrayList<ListenableFuture<T>>();
final Callable<T> callable = new Callable<T>() {
// Some task you want to complete
};
// Submit all your tasks for execution
futures.add(listeningExecutorService.submit(callable));
// ... add as many tasks as you have
futures.add(listeningExecutorService.submit(callable));
// Get a single Future to wait on
final ListenableFuture<List<T>> future = Futures.allAsList(futures);
Futures.addCallback(future, new FutureCallback<List<T>>() {
#Override
public void onSuccess(final List<T> result) {
// Begin other tasks using `result` (the set of results from the first tasks)
}
#Override
public void onFailure(final Throwable t) {
// ...
}
});
This would be helpful if you don't care about waiting around for the second set of tasks to complete, as Futures.addCallback doesn't return anything.
AsyncFunction example:
final ListenableFuture<O> result = Futures.transform(future, new AsyncFunction<List<T>, O>() {
#Override
public ListenableFuture<O> apply(final List<T> input) {
// Begin other tasks using `input` (the set of results from the first tasks)
}
});
This would be beneficial if you want to wait on the resulting ListenableFuture, or potentially add a third set of tasks that need to happen upon completion of the second set.
I have one "Runnable" threads which is initiating few "Callable" threads and I want to display results when all above threads has finished their jobs.
What is the best way to do it?
My code is as follows
Connector.java (Starting Runnable Thread)
public class Connector {
private static void anyFileConnector() {
// Starting searching Thread
ExecutorService executor = Executors.newFixedThreadPool(100);
executor.submit(traverse, executor);
//HERE I WANT MY ALL SEARCH RESULTS/OUTPUT : CURRENTLY IT IS STARTING OTHER THREADS AND NOT SHOWING ME ANY RESULTS BECAUSE NONE OF THEM WAS FINISHED.(IN CONSOLE, I WAS ABLE TO SEE RESULTS FROM ALL THE THREADS
setSearchResult(traverse.getResult());
executor.shutdown();
}
}
Traverse.java (Runnable Thread)
I am using ExecutorCompletionService to handle it...but it didn't create any difference.
:(
public class Traverse implements Runnable {
public void run() {
ExecutorService executor = Executors.newFixedThreadPool(100);
ExecutorCompletionService<List<ResultBean>> taskCompletionService =
new ExecutorCompletionService<List<ResultBean>>(executor);
try (DirectoryStream<Path> stream = Files
.newDirectoryStream(dir)) {
Search newSearch = new Search();
taskCompletionService.submit(newSearch);
}
list.addAll(taskCompletionService.take().get());
}
}
Search.java (Callable Thread)
public class Search implements Callable<List<ResultBean>> {
public List<ResultBean> call() {
synchronized (Search.class) {
// It will return results
return this.search();
}
}
}
Go for CyclicBarrier and you will be able to achieve this.
A cyclic barrier will perform a task as soon as all the threads are done with their work, this is where you can print the en result.
Check this lik for working of CyclicBarrier : http://javarevisited.blogspot.com/2012/07/cyclicbarrier-example-java-5-concurrency-tutorial.html
Easy - all the Callables will return Future objects which you can used to wait and get the result by calling Future.get() in a blocking wait way. So your problem is just a for loop waiting for each future on the callables blockingly.
After that, just aggregate the results to return to client.
The submit method of executor service can return a list of Future objects. What you can do for your case is call isDone() method of these Future objects in a while loop.
Whenever, any future task gets completed this method will return true. You can now call get() method on this to get the value returned by this task. In this way you could get hold of all the future task values without having to wait for any particular task to get complete (since your first future task could have the longest completion time)