This library provides a small and easy wrapper around PHP's PCNTL extension. It allows running of different processes in parallel, with an easy-to-use API.
You can install the package via composer:
composer require spatie/asyncuse Spatie\Async\Pool;
$pool = Pool::create();
foreach ($things as $thing) {
$pool->add(function () use ($thing) {
// Do a thing
})->then(function ($output) {
// Handle success
})->catch(function (Throwable $exception) {
// Handle exception
});
}
$pool->wait();When creating asynchronous processes, you'll get an instance of ParallelProcess returned.
You can add the following event hooks on a process.
$pool
->add(function () {
// ...
})
->then(function ($output) {
// On success, `$output` is returned by the process or callable you passed to the queue.
})
->catch(function ($exception) {
// When an exception is thrown from within a process, it's caught and passed here.
})
->timeout(function () {
// A process took too long to finish.
})
;Instead of using methods on the $pool object, you may also use the async and await helper functions.
use Spatie\Async\Pool;
$pool = Pool::create();
foreach (range(1, 5) as $i) {
$pool[] = async(function () {
usleep(random_int(10, 1000));
return 2;
})->then(function (int $output) {
$this->counter += $output;
});
}
await($pool);If an Exception or Error is thrown from within a child process, it can be caught per process by specifying a callback in the ->catch() method.
$pool
->add(function () {
// ...
})
->catch(function ($exception) {
// Handle the thrown exception for this child process.
})
;If there's no error handler added, the error will be thrown in the parent process when calling await() or $pool->wait().
If the child process would unexpectedly stop without throwing an Throwable,
the output written to stderr will be wrapped and thrown as Spatie\Async\ParallelError in the parent process.
By type hinting the catch functions, you can provide multiple error handlers,
each for individual types of errors.
$pool
->add(function () {
throw new MyException('test');
})
->catch(function (MyException $e) {
// Handle `MyException`
})
->catch(function (OtherException $e) {
// Handle `OtherException`
});Note that as soon as an exception is handled, it won't trigger any other handlers
$pool
->add(function () {
throw new MyException('test');
})
->catch(function (MyException $e) {
// This one is triggerd when `MyException` is thrown
})
->catch(function (Exception $e) {
// This one is not triggerd, even though `MyException` extends `Exception`
});If you need to stop a pool early, because the task it was performing has been completed by one
of the child processes, you can use the $pool->stop() method. This will prevent the
pool from starting any additional processes.
use Spatie\Async\Pool;
$pool = Pool::create();
// Generate 10k processes generating random numbers
for($i = 0; $i < 10000; $i++) {
$pool->add(function() use ($i) {
return rand(0, 100);
})->then(function($output) use ($pool) {
// If one of them randomly picks 100, end the pool early.
if ($output === 100) {
$pool->stop();
}
});
}
$pool->wait();Note that a pool will be rendered useless after being stopped, and a new pool should be created if needed.
Besides using closures, you can also work with a Task.
A Task is useful in situations where you need more setup work in the child process.
Because a child process is always bootstrapped from nothing, chances are you'll want to initialise eg. the dependency container before executing the task.
The Task class makes this easier to do.
use Spatie\Async\Task;
class MyTask extends Task
{
public function configure()
{
// Setup eg. dependency container, load config,...
}
public function run()
{
// Do the real work here.
}
}
// Add the task to the pool
$pool->add(new MyTask());If you want to encapsulate the logic of your task, but don't want to create a full blown Task object,
you may also pass an invokable object to the Pool.
class InvokableClass
{
// ...
public function __invoke()
{
// ...
}
}
$pool->add(new InvokableClass(/* ... */));You're free to create as many pools as you want, each pool has its own queue of processes it will handle.
A pool is configurable by the developer:
use Spatie\Async\Pool;
$pool = Pool::create()
// The maximum amount of processes which can run simultaneously.
->concurrency(20)
// The maximum amount of time a process may take to finish in seconds.
->timeout(15)
// Configure which autoloader sub processes should use.
->autoload(__DIR__ . '/../../vendor/autoload.php')
// Configure how long the loop should sleep before re-checking the process statuses in microseconds.
->sleepTime(50000)
;If the required extensions (pcntl and posix) are not installed in your current PHP runtime, the Pool will automatically fallback to synchronous execution of tasks.
The Pool class has a static method isSupported you can call to check whether your platform is able to run asynchronous processes.
If you're using a Task to run processes, only the run method of those tasks will be called when running in synchronous mode.
When using this package, you're probably wondering what's happening underneath the surface.
We're using the symfony/process component to create and manage child processes in PHP.
By creating child processes on the fly, we're able to execute PHP scripts in parallel.
This parallelism can improve performance significantly when dealing with multiple synchronous tasks,
which don't really need to wait for each other.
By giving these tasks a separate process to run on, the underlying operating system can take care of running them in parallel.
There's a caveat when dynamically spawning processes: you need to make sure that there won't be too many processes at once,
or the application might crash.
The Pool class provided by this package takes care of handling as many processes as you want
by scheduling and running them when it's possible.
That's the part that async() or $pool->add() does. Now let's look at what await() or $pool->wait() does.
When multiple processes are spawned, each can have a separate time to completion. One process might eg. have to wait for a HTTP call, while the other has to process large amounts of data. Sometimes you also have points in your code which have to wait until the result of a process is returned.
This is why we have to wait at a certain point in time: for all processes on a pool to finish, so we can be sure it's safe to continue without accidentally killing the child processes which aren't done yet.
Waiting for all processes is done by using a while loop, which will wait until all processes are finished.
Determining when a process is finished is done by using a listener on the SIGCHLD signal.
This signal is emitted when a child process is finished by the OS kernel.
As of PHP 7.1, there's much better support for listening and handling signals,
making this approach more performant than eg. using process forks or sockets for communication.
You can read more about it here.
When a process is finished, its success event is triggered, which you can hook into with the ->then() function.
Likewise, when a process fails or times out, the loop will update that process' status and move on.
When all processes are finished, the while loop will see that there's nothing more to wait for, and stop.
This is the moment your parent process can continue to execute.
We've written a blog post containing more information about use cases for this package, as well as making comparisons to other asynchronous PHP libraries like ReactPHP and Amp: http://stitcher.io/blog/asynchronous-php.
composer testPlease see CHANGELOG for more information what has changed recently.
Please see CONTRIBUTING for details.
If you discover any security related issues, please email freek@spatie.be instead of using the issue tracker.
You're free to use this package, but if it makes it to your production environment we highly appreciate you sending us a postcard from your hometown, mentioning which of our package(s) you are using.
Our address is: Spatie, Samberstraat 69D, 2060 Antwerp, Belgium.
We publish all received postcards on our company website.
Spatie is a webdesign agency based in Antwerp, Belgium. You'll find an overview of all our open source projects on our website.
Does your business depend on our contributions? Reach out and support us on Patreon. All pledges will be dedicated to allocating workforce on maintenance and new awesome stuff.
The MIT License (MIT). Please see License File for more information.