Squadron_5

	Squadron 5 - Multithreading and worker pools in Dart Offload CPU-bound and long running tasks and give your mobile and Web apps some air!

Highly recommended

With squadron_builder, any Dart class can be automatically turned into a service running on dedicated threads.

Stay focus on implementing your service, and let squadron_builder generate the boilerplate code to run it on native Isolates as well as Web Workers!

Flutter Demo

A demo is available from GitHub: squadron_sample. It provides a Flutter App running on native and browser platforms, showcasing Squadron integration with Flutter. Some of the service workers rely on squadron_builder but others take the manual approach to better understand how things integrate together.

Another demo including code generation is available at GitHub: flutter_mandel_squadron. This sample is a fork from GitHub: flutter_mandel by escamoteur, which originally runs on native platforms only. With Squadron, the sample also runs in Web browsers!

Features

Worker: a base worker class managing a platform thread (Isolate or Web Worker) and the communication between clients and workers.

WorkerPool<W>: a worker pool for W workers. The number of workers is configurable as well as the degree of concurrent workloads distributed to workers in the pool. Tasks posted to the worker pool can be canceled.

Getting Started

Import squadron from your pubspec.yaml file:

dependencies:
   squadron: ^5.0.0

Usage

The basic idea behind Squadron is to wrap a set of service methods in a cross-platform Worker, enabling seamless access to the service API from both Native and Browser platforms.

To implement this pattern, Squadron provides a base WorkerService class to inherit from and implement the functionality to execute in threads, as well as all the plumbing to bootstrap the service in a platform worker and proxy service calls between the app's main event loop and the workers. This approach enables reusing the code in different scenarios: called directly from the app's main event loop, called from unit tests, or wrapped in a native Isolate or in a Web Worker.

The example below implements a SampleService with a synchronous cpu() method and an asynchronous io() method. The service inherits from WorkerService and must implement the operations map. This collection is essentially a dispatcher used to map service command ids with actual command handlers. Squadron uses this map in platform workers to serve worker requests. The command handlers provided in this map are responsible for retrieving arguments from the WorkerRequest message and providing them to the service method.

class SampleService implements WorkerService {
  Future io({required int milliseconds}) =>
      Future.delayed(Duration(milliseconds: milliseconds));

  void cpu({required int milliseconds}) {
    final sw = Stopwatch()..start();
    while (sw.elapsedMilliseconds < milliseconds) {/* cpu */}
  }

  // command IDs
  static const ioCommand = 1;
  static const cpuCommand = 2;

  // command IDs --> command implementations
  @override
  Map<int, CommandHandler> get operations => {
    ioCommand: (WorkerRequest r) => io(milliseconds: r.args[0]),
    cpuCommand: (WorkerRequest r) => cpu(milliseconds: r.args[0]),
  };
}

The rest is a matter of connecting the service class with platform workers. The following code explains step by step how to assemble the service class with workers and worker pools. In practice, this is just boilerplate code that can be generated using Squadron's annotations together with Squadron's codegen companion package squadron_builder. The code generation approach should be prefered; for more details, see Code generation with squadron_builder.

The sample worker class extends Squadron's Worker class and uses the SampleService class as a contract. The resulting worker can then be used in Dart Isolates as well as Web Workers.

class SampleWorker extends Worker implements SampleService {
  SampleWorker(EntryPoint entryPoint, {List args = const []})
      : super(entryPoint, args: args);

  @override
  Future io({required int milliseconds}) =>
      send(SampleService.ioCommand, [milliseconds]);

  @override
  Future cpu({required int milliseconds}) =>
      send(SampleService.cpuCommand, [milliseconds]);
}

Starting with Squadron 3, the implementation of platform worker's main program is simplified with the run() function. This function takes two arguments:

first, a WorkerService initializer responsible for creating the service to be used by the platform worker.
second, a WorkerRequest that will enable setting up the service; this argument is not used for Browser platform but it is required in native scenarios where it must be set to the Isolate's main program parameter.

Here is the code for:

a native worker running in an Isolate:

SampleWorker createVmSampleWorker() => SampleWorker(_main);

// Isolate entry-point.
// It must be a top level function or static method accepting a Map (Squadron < 5.0.0) or a List (Squadron >= 5.0.0) argument.
// The argument passed to the entry-point must be passed to the run() function.
void _main(List command) => run((startRequest) => SampleService(), command);

a browser worker running in a Web Worker:

SampleWorker createJsSampleWorker() => SampleWorker('sample_worker_js.dart.js');

// Web Worker main program.
// It must be a parameter-less "main()" function.
void main() => run((startRequest) => SampleService());

Using a WorkerPool, you are now able to distribute your workloads:

    var pool = WorkerPool(createVmSampleWorker, maxWorkers: 4, maxParallel: 2); /* native version */
    // var pool = WorkerPool(createJsSampleWorker, maxWorkers: 4, maxParallel: 2); /* browser version */
    await pool.start();

    var n = 42;
    var cpuResult = await pool.execute((w) => w.cpu(milliseconds: n));
    var ioResult = await pool.execute((w) => w.io(milliseconds: n));

Code generation with squadron_builder

To simplify the developer experience, Squadron provides annotations leveraged by squadron_builder to automatically generate the code for workers and worker pools. As a result, developers need to focus on functionality only, leaving plumbing and bootstrapping aspects to squadron_builder.

part 'sample_service.worker.g.dart';

@SquadronService()
class SampleService {
  @SquadronMethod()
  Future io({required int milliseconds}) =>
      Future.delayed(Duration(milliseconds: milliseconds));

  @SquadronMethod()
  Future cpu({required int milliseconds}) async {
    final sw = Stopwatch()..start();
    while (sw.elapsedMilliseconds < milliseconds) {/* cpu */}
  }
}

Remarks on Isolates / Web Workers

While Isolates enable multithreading in Dart applications, several aspects must be taken into account:

Creating an Isolate can be an expensive process.
Isolate threads are still based on an event loop, processing asynchronous tasks/callbacks one by one.
Isolates do not share memory; in particular, global contexts are not shared across Isolates and may have to be initialized multiple times thus increasing the application's memory footprint and startup time.
Communicating with an Isolate invoves marshaling data in and out; theoretically, only primitive types (num, String, bool, null...), List/Map of primitive types and some specific types like SendPort are supported. However, most object instances may be sent across Isolates on native platforms (instances will still be copied).

Web Workers have similar characteristics but only primitive types and objects implementing Transferable can be sent across Web Worker boundaries.

Communication across threads does not come for free, and some experiments have found that using jsonEncode() may actually be more efficient when sending large data sets (e.g. a List containing many Maps) even with the overhead of calling jsonDecode() on the receiving end. This is especially true on Web platforms because Dart's implementation of postMessage involves converting Dart objects to native JavaScript objects. These objects will then be cloned by the browser.

Channels, Types, and Browser Platforms

To provide a cross-platform development experience, Squadron encapsulates Isolates and Web Workers as well as the means to communicate between the main app's code and the code they execute. This is achieved via the Channel class.

Diagram about type flow through Squadron Channels

Channel enables data exchange between threads and inherits the constraints of the target platforms, in particular the type system. Dart Native platforms will typically be quite relaxed when communicating between threads, even allowing custom Dart objects to come through.

But JavaScript will not be so forgiving because JavaScript doesn't know about Dart types.

To transfer a custom object, it must be serialized on the sender end and deserialized on the receiver end. There are several ways of serializing a custom object, e.g. JSON structure holding the object's attributes, or String/binary representation of the object...

However, when the data to be transfered hits the browser's Web Worker implementation, only basic type information (number, boolean, string, array or map) is retained. In particular, generic types sent from one side will not be received with the same generic type on the other end. For instance, when a sending a List<String> or a Map<String, dynamic> to a service worker, browser platforms will provide the data to the worker service as a bare List (= List<dynamic>) or a bare Map (= Map<dynamic, dynamic>). This is an important point to ensure your app will happily run on browsers.

Serialization and deserialization should be done as close to the Channel as possible, typically when calling the send method or when receiving data in the operations map.

For example, suppose we have the following service definition, processing ServiceRequest messages from callers and replying with ServiceResponse objects:

abstract class ServiceDefinition {
   FutureOr<ServiceResponse> serviceMethod(ServiceRequest request);

   static const cmdServiceMethod = 1;
}

class ServiceResponse {
   static ServiceResponse deserialize(dynamic result) {
      // deserialize "result", knowing it was produced by serialize()
      // or call the code produced by package:json_annotation with anyMap = true
   }

   dynamic serialize() {
      // serialize using only base types or simple List/Map
      // or call the code produced by package:json_annotation with anyMap = true
   }
}

class ServiceRequest {
   static ServiceRequest deserialize(dynamic result) {
      // deserialize "result", knowing it was produced by serialize()
      // or call the code produced by package:json_annotation with anyMap = true
   }

   dynamic serialize() {
      // serialize using only base types or simple List/Map
      // or call the code produced by package:json_annotation with anyMap = true
   }
}

Then the approach below is a generic way to send/receive the ServiceRequest/ServiceResponse messages via Squadron:

class ServiceImplementation implements ServiceDefinition {
   @override
   ServiceResponse serviceMethod(ServiceRequest request) {
      // process the request and produce the response
   }

   // in the operations map, deserialize argument and serialize result

   @override
   late final Map<int, CommandHandler> operations = {
      ServiceDefinition.cmdServiceMethod: (WorkerRequest r) =>
          serviceMethod(
            ServiceRequest.deserialize(r.args[0])   // deserialize ServiceRequest
          ).serialize();                            // serialize ServiceResponse
   };
}

class ServiceWorker extends Worker implements ServiceDefinition {

   // in the worker overrides, serialize argument and deserialize result

   @override
   ServiceResponse serviceMethod(ServiceRequest request) async =>
        ServiceResponse.deserialize(                // deserialize ServiceResponse
          await send(
            ServiceDefinition.cmdServiceMethod,
            [ request.serialize() ]                 // serialize ServiceRequest
          )
        );
}

class ServiceWorkerPool extends WorkerPool<ServiceWorker> implements ServiceDefinition {
  ServiceWorkerPool(EntryPoint entryPoint, ConcurrencySettings concurrencySettings)
      : super(() => ServiceWorker(entryPoint),
            concurrencySettings: concurrencySettings);

   // nothing to do in the service pool, the service worker and service implementation
   // take care of all serialization aspects

  @override
  Future<ServiceResponse> serviceMethod(ServiceRequest request) =>
      execute((w) => w.serviceMethod(request));
}

Marshaling data with `squadron_builder`

Serialization/deserialization aspects can be delegated to squadron_builder in several ways:

custom types may be decorated with @Marshaler where Marshaler is the class responsible for marshaling/unmarshaling class instances through platform channels. Marshalers can also be used to annotate a method or a parameter to explicitly set another marshaler. Marshaler must be a class (with a const constructor) deriving from SquadronMarshaler<T, S>, or a constant instance of such a class, where T is the type of the service method return value or parameter and S the type used for transit. For instance, data could be transported as bytes like in the following example:

@SquadronService()
class DataService {

  @SquadronMethod()
  // will use DataMarshaler by default
  FutureOr<Data> doSomething(Data input) =>
      Data(input.a + 1, !input.b, '(was $input)');

  @SquadronMethod()
  @SpecificDataMarshaler()
  // will use DataMarshaler for the `input` parameter, and SpecificDataMarshaler for the return value
  FutureOr<Data> doSomethingElse(Data input) =>
      Data(input.a + 1, !input.b, '(was $input)');

  @SquadronMethod()
  // will use DataMarshaler for the return value, and SpecificDataMarshaler for the `input` parameter
  FutureOr<Data> doAnotherThing(@SpecificDataMarshaler() Data input) =>
      Data(input.a + 1, !input.b, '(was $input)');
}

@DataMarshaler()
class Data {
  Data(this.a, this.b, this.c);

  final int a; // 32-bits
  final bool b;
  final String c;

  @override
  String toString() => 'a=$a, b=$b, c=$c';
}

class DataMarshaler extends SquadronMarshaler<Data, Uint8List> {
  const DataMarshaler();

  @override
  Uint8List marshal(Data payload) {
    final str = utf8.encode(payload.c);
    final buffer = Uint8List(4 + 1 + str.length);
    _writeInt(buffer, 0, payload.a);
    buffer[4] = payload.b ?  1 : 0;
    buffer.setRange(5 /* = 4 + 1 */, buffer.length, str);
    return buffer;
  }

  @override
  Data unmarshal(Uint8List buffer) {
    final a = _readInt(buffer, 0);
    final b = (buffer[4] != 0);
    final c = utf8.decode(buffer.skip(5).toList());
    return Data(a, b, c);
  }

  static void _writeInt(Uint8List buffer, int pos, int value) {
    buffer[pos  ] = (value & 0xFF000000)>>24;
    buffer[pos+1] = (value & 0x00FF0000)>>16;
    buffer[pos+2] = (value & 0x0000FF00)>>8;
    buffer[pos+3] = (value & 0x000000FF);
  }

  static int _readInt(Uint8List buffer, int pos) =>
        (buffer[pos]<<24) | (buffer[pos+1]<<16) 
      | (buffer[pos+2]<<8) | buffer[pos+3];
}

class SpecificDataMarshaler extends SquadronMarshaler<Data, Uint8List> {
  const SpecificDataMarshaler();

  /* alternative implementation */
}

return value/parameter classes may also implement their own marshal()/unmarshal() methods. If no @SerializeWith() annotation is present, these methods will be used for marshaling/unmarshaling data through platform channels, like in the following example:

@SquadronService()
class DataService {

  @SquadronMethod()
  FutureOr<Data> doSomething(Data input) =>
      Data(input.a + 1, !input.b, '(was $input)');
}

class Data {
  Data(this.a, this.b, this.c);

  final int a; // 32-bits
  final bool b;
  final String c;

  @override
  String toString() => 'a=$a, b=$b, c=$c';

  Uint8List marshal() {
    final str = utf8.encode(c);
    final buffer = Uint8List(4 + 1 + str.length);
    _writeInt(buffer, 0, a);
    buffer[4] = b ?  1 : 0;
    buffer.setRange(5 /* = 4 + 1 */, buffer.length, str);
    return buffer;
  }

  Data.unmarshal(Uint8List buffer)
      : a = _readInt(buffer, 0),
        b = (buffer[4] != 0),
        c = utf8.decode(buffer.skip(5).toList());

  static void _writeInt(Uint8List buffer, int pos, int value) {
    buffer[pos  ] = (value & 0xFF000000)>>24;
    buffer[pos+1] = (value & 0x00FF0000)>>16;
    buffer[pos+2] = (value & 0x0000FF00)>>8;
    buffer[pos+3] = (value & 0x000000FF);
  }

  static int _readInt(Uint8List buffer, int pos) =>
        (buffer[pos]<<24) | (buffer[pos+1]<<16) 
      | (buffer[pos+2]<<8) | buffer[pos+3];
}

alternatively, squadron_builder will consider toJson()/fromJson() methods (which could be generated eg. with json_serializable). When applicable, custom-typed members must also be taken into account and serialized/deserialized.

@SquadronService()
class DataService {

  @SquadronMethod()
  FutureOr<Data> doSomething(Data input) =>
      Data(input.a + 1, !input.b, '(was $input)');
}

class Data {
  Data(this.a, this.b, this.c);

  final int a; // 32-bits
  final bool b;
  final String c;

  @override
  String toString() => 'a=$a, b=$b, c=$c';

  Map toJson() => {'a': a, 'b': b, 'c': c};

  static Data fromJson(Map json) => Data(json['a'], json['b'], json['c']);
}

These approaches must comply with the following invariants to ensure that data sent through a channel is received on the other side as a clone that is functionally equivalent to the original data:

with marshaler M: the output of M.marshal(M.unmarshal(data)) must be functionally equivalent to the original data.
with marshal()/unmarshal() methods: the output of Data.unmarshal(data.marshal()) must be functionally equivalent to the original data.
with toJson()/fromJson() methods: the output of Data.fromJson(data.toJson()) must be functionally equivalent to the original data.

Note on `package:json_annotation`

Packages such as json_annotation / json_serializable can be used to generate the serialization and deserialization code for custom classes. By default, objects will be serialized to / deserialized from Map<String, dynamic> data structures.

In Browser scenarios, this will lead to errors as the Map<String, dynamic> structures lose their strong types when they get processed by the browser.

Luckily, json_annotation provides the anyMap option to control code generation: by setting anyMap to true, the code builders from json_serializable will handle JSON objects as Map instead of Map<String, dynamic>.

On Web platforms, setting anyMap to true is mandatory for classes that are transferred to/from Squadron workers.

Scaling Options

Squadron pools manage a collection of workers to avoid the cost of creating a new platform worker each time. Squadron also implements a simple load-balancing mechanism and posts new tasks to workers that are most available (i.e. those with the minimum workload, provided workload < maxParallel). Tasks in Squadron do not have a priority and will be handled in the order they were received.

Depending on the type of processing, scaling tasks with Squadron can be achieved horizontally (adding new workers) or vertically (distributing more tasks to workers).

For pure CPU-bound tasks, e.g. image/video or other heavy data processing, increasing the maxParallel pool option is likely to NOT yield any benefit for performance. If the event loop is already busy executing a task, subsequent task requests posted to the Worker will not be handled before the executing task is complete. As a result, CPU-bound tasks will be queued and scaling CPU workloads can only be achieved by adding more workers to the pool (horizontal scaling).
For I/O bound tasks, performance benefits are less obvious. However, if processing requires many I/O operations, offloading such tasks to a worker pool is likely to be beneficial because I/O callbacks will be registered with the platform worker's event loop. This scenario can be interesting in Web application back-ends to make the main event loop more available for other incoming requests -- simple requests e.g. CRUD operations can be fully handled by the main event loop while more complex, long-running, I/O-bound tasks will be processed off the Web app's main event loop. Scaling such tasks can be achieved by increasing the maxParallel (vertical scaling) or maxWorkers (horizontal scaling) pool options. In I/O scenarios, vertical scaling should be preferred.

Worker Cooperation

It is possible to implement some kind of worker cooperation and support more complex scenarios.

For instance, the fact that Isolates and Web Workers do not share memory means it may be cumbersome to implement a local, in-memory cache at worker level. Each worker would have their own cache, making expiration and update propagation difficult to implement.

Such a cache can however be implemented as a Singleton, communicating with other workers by sharing its Channel property (Channel objects can be sent across platform workers via the serialize() method).

An example is provided in cache_worker.dart. To access the cache API seamlessly, an abstract class is first defined:

abstract class Cache {
  FutureOr<dynamic> get(dynamic key);
  FutureOr<bool> containsKey(dynamic key);
  FutureOr set(dynamic key, dynamic value, {Duration? timeToLive});
  FutureOr<CacheStat> getStats();
}

A cache service is then implemented (internal details skipped, please refer to the example source code). Note that the CacheService implementation can be synchronous.

class CacheService implements Cache, WorkerService {
  @override
  dynamic get(dynamic key) {
    // retrieve the value associated for the specified key
    // return null if key is not in cache or if the key has expired
  }

  @override
  bool containsKey(dynamic key) {
    // use get() as it implements the expiration logic
    return get(key) != null;
  }

  @override
  void set(dynamic key, dynamic value, {Duration? timeToLive}) {
    // cache the value with the specified key and TTL
  }

  @override
  CacheStat getStats() {
    // return cache stats
  }

  // command IDs
  static const getOperation = 1;
  static const containsOperation = 2;
  static const setOperation = 3;
  static const statsOperation = 4;

  // command handlers
  Map<int, CommandHandler> get operations => {
        getOperation: (WorkerRequest r) => get(r.args[0]),
        containsOperation: (WorkerRequest r) => containsKey(r.args[0]),
        setOperation: (WorkerRequest r) => set(r.args[0], r.args[1],
            timeToLive: (r.args[2] == null) ? null : Duration(microseconds: r.args[2])),
        statsOperation: (WorkerRequest r) => getStats().serialize()
      };
}

The CacheWorker is easy:

class CacheWorker extends Worker implements Cache {
  CacheWorker(EntryPoint entryPoint, {List args = const []})
      : super(entryPoint, args: args);

  @override
  Future<dynamic> get(dynamic key) => send(CacheService.getOperation, [key]);

  @override
  Future<bool> containsKey(dynamic key) =>
      send(CacheService.containsOperation, [key]);

  @override
  Future set(dynamic key, dynamic value, {Duration? timeToLive}) {
    assert(value != null); // null means not in cache; cannot store null
    return send(CacheService.setOperation, [key, value, timeToLive?.inMicroseconds]);
  }

  @override
  Future<CacheStat> getStats() async =>
      CacheStat.deserialize(await send(CacheService.statsOperation));
}

Note how getStats() implementations require serialization/deserialization of the CacheStat object. This is necessary to cross platform worker boundaries.

Finally, a cache client is implemented to proxy calls from other workers to the CacheWorker. This CacheClient can be constructed with a Channel that will be obtained from the CacheWorker.

class CacheClient implements Cache {
  CacheClient(this._remote);

  final Channel _remote;

  @override
  Future<dynamic> get(dynamic key) =>
      _remote.sendRequest(CacheService.getOperation, [key]);

  @override
  Future<bool> containsKey(dynamic key) =>
      _remote.sendRequest(CacheService.containsOperation, [key]);

  @override
  Future set(dynamic key, dynamic value, {Duration? timeToLive}) {
    assert(value != null); // null means not in cache; cannot store null
    return _remote.sendRequest(
        CacheService.setOperation, [key, value, timeToLive?.inMicroseconds]);
  }

  @override
  Future<CacheStat> getStats() async {
      => CacheStat.deserialize(
        await _remote.sendRequest(CacheService.statsOperation, []));
}

The following service is an example for a computation leveraging the shared cache. Its constructor takes a Cache parameter.

class OtherService implements WorkerService {
  OtherService([this._cache]);

  final Cache? _cache;

  // some service method
  Future<int> compute(int n) async {
    // check cache
    int? result = await _cache?.get(n);
    if (result != null) {
      // cache hit
      return n;
    }
    // otherwise compute
    // ...
    // finally, cache (dont await) and return
    cache?.set(n, result);
    return result!;
  }

  // some command ids
  static const computeCommand = 1;

  @override
  Map<int, CommandHandler> get operations => {
    // the comand handlers for the command ids
    computeCommand: (r) => compute(r.args[0])
  };
}

The OtherWorker implementation is straightforward:

class OtherWorker extends Worker implements OtherService {
  OtherWorker(EntryPoint entryPoint, {List args = const []})
      : super(entryPoint, args: args);

  @override
  Future<int> compute(int n) => send<int>(OtherService.computeCommand, [n]);

  // other proxy service methods...

  @override
  final Cache? _cache = null;
}

The platform worker assembles everything. It is essentially the same as above, with some extra initialization code to set up a CacheClient and provide it to the CacheService.

To create a CacheClient from within the platform worker, the CacheWorker's Channel must be passed to the OtherWorker. This is done using the share() and the serialize() methods provided by Channel. These methods will return an opaque object that can be safely sent across workers and deserialized to recreate a Channel, thereby bridging the gap between the OtherService instances and the CacheService Singleton.

OtherWorker createOtherWorker([CacheWorker? cache]) =>
    OtherWorker(_main, args: [cache?.channel?.share().serialize()]);

void _main(Map command) {
  run((startRequest) {
    final cacheEndPoint = startRequest.args.isEmpty
        ? null
        : Channel.deserialize(startRequest.args[0]);
    Cache? cache = (cacheEndPoint == null) ? null : CacheClient(cacheEndPoint);
    return OtherService(cache);
  }, command);
}

The application's main program is responsible for:

setting up the CacheWorker Singleton
setting up a pool of OtherWorkers and making sure the worker factory function receives the CacheWorker Singleton.

final cacheWorker = CacheWorker();
await cacheWorker.start();

final pool = WorkerPool(() => createOtherWorker(cacheWorker), minWorkers: 2, maxWorkers: 5);
await pool.start();

Architecture Diagram

                                   +-----------------------+
                              +--> | CacheWorker singleton | <--+
                              |    +-----------------------+    |
                              |                                 |
 +-------------------+        |           +----------------+   [4]
 |  main program     | --[1]--+    +----> | otherWorker #1 |    |
 |  ---------------- |             |      | -------------- |    |
[2] OtherWorker pool | --[3]--+    |      | cacheClient #1 | ---|
 +-------------------+        |    |      +----------------+    |
                              +----|                            |
                                   |      +----------------+    |
                                   +----> | otherWorker #2 |    |
                                          | -------------- |    |
                                          | cacheClient #2 | ---+
                                          +----------------+

1: the main program first spawns a CacheWorker
2: the main program creates a pool of OtherWorkers, making sure the CacheWorker Singleton is advertized to OtherWorkers when they are created
3: the pool creates new OtherWorkers as workload builds up, instantiating cache clients bound to the CacheWorker's Channel
4: OtherWorkers query the CacheWorker via their local CacheClient to avoid expensive computations that have been done already

Starting with Squadron 3.3, the design for a shared cache could be based on a LocalWorker for the CacheWorker. However if the cache is often used from workers, it would increase the load on the main thread's event loop and may impact the user experience.

Local Workers

Squadron 3.3 introduces LocalWorker, a Woker-like class running in the same thread as its owner. The main idea behind LocalWorker is to enable executing code in the context of the owner thread such as a the main thread in a Flutter app, thereby giving access to Flutter APIs. In this scenario, it is important to note that the LocalWorker will use the event loop of the main app. As a result, if the load on the LocalWorker is high enough, it could impact the responsiveness of the application.

The implementation of a LocalWorker follows the same principles as for a Worker. Start by implementing a WorkerService with the logic you want to expose:

// The identity service
@SquadronService()
class IdentityService {
  @squadronMethod
  FutureOr<int> whoAreYou() => Isolate.current.hashCode;
}

To allow other workers to communicate with the LocalWorker, a LocalWorkerClient must be implemented. The client will be passed a Channel obtained from the LocalWorker, and simply forwards commands to the LocalWorker.

// The service client: this class will be used in workers that need to call the service implementation
class IdentityClient extends LocalWorkerClient implements IdentityService {
  IdentityClient(Channel channel) : super(channel);

  @override
  Future<int> whoAreYou() =>
      send(IdentityService.whoAreYouCommand, []);
}

Next, we have the code for the real Worker that will be using the LocalWorker. Typically, it is based on a WorkerService, for instance:

@SquadronService()
class MyWorkerService{
  MyWorkerService(this._identityClient);

  final IdentityClient _identityClient;

  @squadronMethod
  Future<String> whoAreYouTalkingTo() async {
    // this is where the local worker is called
    final localWorkerIdentity = await _identityClient.whoAreYou();
    return 'I am running in ${Isolate.current.hashCode}, and I am talking to $localWorkerIdentity.';
  }
}

The code to fire up the workers in browser and VM worlds need to get the Channel from the LocalWorker, so we will pass the local worker to the functions that instantiate MyWorker. The channel is obtained by calling the share() method then passed on to MyWorker via the args parameter where it must be serialized to cross platform thread boundaries. This array will be received by the platform worker thread at startup via the startRequest argument:

And that's it! The final step is to bind everything together, for instance in the main program:

void main() {
  Squadron.setId('main');

  final identityServer = LocalWorker<IdentityService>.create(IdentityServiceImpl());

  print(identityServer.whoAreYou());

  try {
    final worker = createMyWorker(identityServer);
    print(await worker.whoAreYouTalkingTo());
  } finally {
    identityServer.stop();
  }
}

Note: the LocalWorker must be stopped to stop the program.

Architecture Diagram

                                  +----------------+
                             +--> | MyWorker       |
                             |    | -------------- |
                             |    | identityClient | --+
 +------------------+        |    +----------------+   |
 |  main program    | --[2]--+                        [3]
 |  --------------- |                                  |
[1] identityServer  | <--------------------------------+
 +------------------+ 

1: the main program first creates a LocalWorker<IdentityService>
2: the main program spawns a MyWorker, providing it with the LocalWorker<IdentityService> it has just created
3: MyWorker will call into the LocalWorker running in the main thread via its LocalWorkerClient

Task Cancelation

Tasks registered with the worker pool may be canceled by calling pool.cancel(). A CanceledException will be raised (for value tasks: the future completes with an error) or emitted (for streaming tasks: the stream will emit an error) for each canceled task. Tasks still pending will fail immediately; tasks already executing when the cancel() method is called will either complete (value task) or emit an exception (streaming tasks).

It should be noted that implementations relying on pool.cancel() will not notify platform workers about the cancelation. Tasks that have been assigned to a platform worker will continue executing until they complete. As a result, a value task already executing cannot be canceled this way: it will complete and return a value. The situation is slightly different for a streaming task: while it will report cancelation in the main event loop, streaming will continue in the platform worker's event loop.

  final future = pool.execute((w) => w.computeData());
  final stream = pool.stream((w) => w.streamData());

  // no async suspension means Squadron did not have any chance to schedule any task
  // as a result this cancelation request will cancel both tasks
  pool.cancel();

  stream.listen(
    (value) => print('received value: $value'),   // will not be called
    onError: (e) => print('received error: $e')); // receives a CanceledException

  final result = await future; // throws a CanceledException

  final future = pool.execute((w) => w.computeData());
  final stream = pool.stream((w) => w.streamData());

  // asynchronous suspension gives Squadron a chance to schedule some tasks
  Future(() => null);

  // depending on concurrency settings, this cancelation request should cancel the stream task
  // however the value task should have been scheduled and should complete
  pool.cancel(); 

  stream.listen(
    (value) => print('received value: $value'),   // may be called for a few values
    onError: (e) => print('received error: $e')); // will receive a CanceledException

  final result = await future; // should get the task's result

It is also possible to schedule and cancel individual tasks, eg.:

  final valueTask = pool.scheduleTask((w) => w.computeData());
  final streamTask = pool.scheduleStream((w) => w.streamData());

  // no async suspension means Squadron could not schedule any task

  streamTask.cancel(); // or pool.cancel(streamTask)
  streamTask.stream.listen(
    (value) => print('received value: $value'),   // will not be called
    onError: (e) => print('received error: $e')); // receives a CanceledException

  valueTask.cancel(); // or pool.cancel(valueTask)
  final result = await valueTask.value; // throws a CanceledException

  final valueTask = pool.scheduleTask((w) => w.computeData());
  final streamTask = pool.scheduleStream((w) => w.streamData());

  // no async suspension means Squadron did not have any chance to schedule any task
  Future(() => null);

  streamTask.cancel(); // or pool.cancel(streamTask)
  streamTask.stream.listen(
    (value) => print('received value: $value'),   // may be called for a few values
    onError: (e) => print('received error: $e')); // will receive a CanceledException

  valueTask.cancel(); // or pool.cancel(valueTask)
  final result = await valueTask.value;  // should get the task's result

Cancelation Tokens

Squadron leverages package [cancelation_token][pub_cancelation_token] for cancelation tokens. To ensure workers are notified of a token's cancelation, the token must be provided to the worker. Cancelation notifications are posted to workers regardless of the maxParallel concurrency settings. It is the responsibility of the worker service (your code) to verify the token status and abort processing when requested. To ensure the service method can process the cancelation request, it must be asynchronous.

When a token is canceled, all tasks associated with the token will be canceled. A CanceledException will be thrown and the code that started the worker task will receive it.

Example:

class SampleService implements WorkerService {
  Future io({required int milliseconds, CancelationToken? token}) async {
    await token.throwIfCanceled();
    await Future.delayed(Duration(milliseconds: milliseconds));
  }

  Future cpu({required int milliseconds, CancelationToken? token}) async {
    final sw = Stopwatch()..start();
    var count = 0;
    while (sw.elapsedMilliseconds < milliseconds) {
      if (count % 1000 == 0) {
        // avoid flooding the event loop with noop futures, check every 1000 iterations only
        await token.throwIfCanceled();
      }
      count++;
    }
  }

  // command ids
  static const ioCommand = 1;
  static const cpuCommand = 2;

  // map of command ids to implementatons
  @override
  Map<int, CommandHandler> get operations => {
    ioCommand: (WorkerRequest r) => io(milliseconds: r.args[0], token: r.cancelToken),
    cpuCommand: (WorkerRequest r) => cpu(milliseconds: r.args[0], token: r.cancelToken),
  };
}

  // create a token
  final token = CancelableToken();
  // trigger cancelation after 500 ms
  // in real world, token.cancel() would be called in reaction to an event such as a user action for instance
  // this is similar to a timeout token except that countdown starts immediately
  Timer(Duration(milliseconds: 500), token.cancel);

  // start a computation lasting 1000 ms
  // pass the token to the service + the pool
  // a CanceledException will be thrown after 500 ms
  await pool.execute((w) => w.cpu(milliseconds: 1000, token));

Notes:

the token received by workers will not have the same runtime type as the token passed to the worker. The service code should not make any assumption on the token's runtime type and should only manipulate generic CancelationToken objects.
the same cancelation token may be used to control cancelation of several tasks.

Worker Monitoring

Monitoring workers in a pool can be done with a simple timer. For instance, to stop workers after a given idle period:

  // install worker monitor
  final timer = Timer.periodic(refreshDuration, (timer) {
    pool.stop((w) => w.idleTime > maxIdle);
  });

Please note that some idle workers may remain alive, depending on the minWorker concurrency setting.

To stop all workers, simply call the pool's stop() method with no predicate.

  // stops all workers
  pool.stop();

All workers will be stopped as soon as all tasks registered with the pool have been processed. Of course, it is possible to cancel pending tasks before stopping the worker pool.

  // cancels pending tasks and stops all workers
  pool.cancel();
  pool.stop();

The pool will not accept new tasks unless it is restarted with pool.start().

Logging

Squadron leverages package [logger][pub_logger] for logging. Worker and WorkerPool provide a channelLogger field which may be set by the main program after creating a worker or a worker pool. Log messages emitted from a platform thread will be redirected to this logger if set.

Releasing Your App

Releasing your application for Dart Native platforms should be straightforward.

To release the Browser version, the worker code must first be compiled to JavaScript:

dart compile js lib/src/browser/service_worker.dart -o web/service_worker.dart.js

Flutter's Web runtime includes a service worker called flutter_service_worker.js: make sure you use a different name for your workers to avoid any conflict! After compiling your worker scripts, you can build your app using:

flutter build web

Thanks!

Saad Ardati for his patience and feedback when implementing Squadron into his Flutter application.
Martin Fink for the feedback on Squadron's Stream implementation -- this has resulted in huge progress and a major improvement!
Klemen Tusar for providing a sample Chopper JSON decoder leveraging Squadron.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Squadron_5

Squadron 5 - Multithreading and worker pools in Dart

Table of Contents

Highly recommended

Flutter Demo

Features

Getting Started

Usage

Code generation with squadron_builder

Remarks on Isolates / Web Workers

Channels, Types, and Browser Platforms

Marshaling data with `squadron_builder`

Note on `package:json_annotation`

Scaling Options

Worker Cooperation

Local Workers

Task Cancelation

Cancelation Tokens

Worker Monitoring

Logging

Releasing Your App

Thanks!

Clone this wiki locally

Squadron_5

Squadron 5 - Multithreading and worker pools in Dart

Table of Contents

Highly recommended

Flutter Demo

Features

Getting Started

Usage

Code generation with squadron_builder

Remarks on Isolates / Web Workers

Channels, Types, and Browser Platforms

Marshaling data with squadron_builder

Note on package:json_annotation

Scaling Options

Worker Cooperation

Local Workers

Task Cancelation

Cancelation Tokens

Worker Monitoring

Logging

Releasing Your App

Thanks!

Clone this wiki locally

Marshaling data with `squadron_builder`

Note on `package:json_annotation`