thread-task

What is thread-task? In essence, it is a multithreaded task library for web applications. "But web applications are single-threaded," you say! Indeed they are. But modern browsers implement something called a web worker. The issue with using web workers is that the API is silly and workers in general are very limited in that they can only pass very basic databack and forth to each other. Using thread-task, you get a Task object that is much closer to something like Task from the .NET runtime: you provide a function that represents a task that needs to be run in the background on another thread, and when that task finishes, it gives the result back to your main application. But it does so in a way that lets you send complex objects and data structures -- much more complex than could be done using plain web workers. It also provides a nice Pipe class that allows you to communicate between your worker thread and your main application thread in a way that is simple and understandable.

Another key difference is that unlike with a web worker, you can pass complex argument objects to your Task and it will have access to those arguments when it executes your task.

Putting all of this together, it allows you to move expensive operations into a concurrent thread in a way that is dead simple, and in a way that lets you communicate using your real state objects, no matter how complex they are!

Note also that the library supports having long-running background threads. You can make use of such threads using the Pipe system to transmit complex data types back and forth from your background thread and your primary application thread. When you have some work that needs to be done, just pipe it over and wait for the response.

Usage

Let's look first at the simplest, most contrived possible use of Task:

  const {promise} = Task.run(() => {
    console.log('I am doing this from my task!');
    return 1;
  });

  promise.then(taskResult => {
    console.log('The result from the background task was: ', taskResult);
  });

This will produce this output in the browser console:

I am doing this from my task!
The result from the background task was: 1

This is not all that useful. First off, it is not a realistic use-case because we are not passing any data to the task. It's unlikely that your application will be able to run background operations without any context or arguments. So let's get a bit more creative with this, and create a task that accepts a multiply function that will multiply the value it is given by 10. But this logic originates in the main application and is provided to your task as an argument.

  const myTask = (multiply: (value: number) => number) => multiply(500);

  const result = await Task.run(myTask,
    function (v) { // multiply function
      return v * 2;
    });

  console.log('Task returned: ', result);

This would produce this message in this console:

Task returned: 1000

Or, let's say we have an even more complex operation that first has to request some data from a server and then process it inside of the worker thread:

  const myTask = (dataUri: string) => {
    return fetch(dataUri)
      .then(r => r.json())
      .then(j => {
        return j.map(v => {
          for (let j = 0; j < 10000; ++j) {
            // simulate a CPU-hog operation
          }

          return v * 10;
        });
      });
  };

  const taskResult = await Task.run(myTask, 'http://foobar.com/test-data');

  console.log('Result: ', taskResult);

If our service were returning a JSON document of [0, 1, 2], this would produce in the console:

Result: [0, 10, 20]

Communication

So far so good. What if we have an even more complex use case where we want to pass data to and from the Task while it is running in the background? No problem; thread-task includes the concept of a Pipe that allows you to do this. Note that while web workers already have a message-passing API that you can use without thread-task, that message-passing system only allows you to transfer extremely simple JSON objects. If your object has a circular reference, or a prototype other than Object.prototype, or if it contains some instance method functions, you cannot pass these to a web worker without task-thread. With task-thread, you can pass objects of absolutely any complexity through:

• A communication Pipe object
• Task argument list
• Task return value / result

The only limitation here is that if you are passing functions to a task or from a task, it will work, but your functions absolutely cannot reference closure variables of any kind unless you know for a fact that they will be available in the separate worker thread context (which they almost never will be). So you can pass a function, but do not rely on closure values inside that function if you expect to be passing it to a Task or receiving it from a Task.

Communication example

This example nicely illustrates the library's ability to transmit complex values back and forth, including class instances, through the pipe system. This would be impossible just using plain web workers without thread-task. This code is taken from the Task unit tests in task.test.ts:

  it('should be able to pass a complex class instance through a Pipe and call one of its methods',
      done => {
        const task = Task.run(pipe => {
          class MyClass {
            private instanceVariable = 10;

            public myFunction() {
              return `My instance variable is ${this.instanceVariable}`;
            }
          }

          pipe.postMessage(new MyClass());
        });

        task.pipe.subscribe(
          myClassInstance => {
            const returnValue = myClassInstance.myFunction();
            expect(returnValue).toBe('My instance variable is 10');
            done();
          });
      });

Or this test, which does something similar but passes the structure in the opposite direction, modifies the structure inside the task, and passes it back to the application:

  it('should be able to give a complex data type to a Task and get it back in a return value',
    done => {
      class UselessDataStructure {
        private elements = new Set<string>();

        add(value: string) {
          this.elements.add(value);
        }

        size(): number {
          return this.elements.size;
        }
      }

      const run = async () => {
        const structure = new UselessDataStructure();

        const fromTask = await Task.run<UselessDataStructure>(
          struct => {
            struct.add('a');
            struct.add('b');
            struct.add('c');
            return struct;
          },
          structure).promise;

        expect(fromTask.size()).toBe(3);

        done();
      };

      run();
    });