jordalgo-task

Task

A javascript data type for async requests. Very similar to the data.task and fun-task with some modifications. Published as jordalgo-task

Installing

npm install jordalgo-task

Example

import Task from 'Task';
const task = new Task((sendFail, sendSuccess) => {
  const id = setTimeout(() => {
    sendSuccess(1);
  }, 1000);
  return () => { clearTimeout(id); };
});
 
const cancel = task.run(
  fail => {
    // never called;
  },
  success => {
    // success === 1
  }
);

Quick Details

• The functions passed to run are always called async
• You can't complete a Task more than once e.g. you can't call sendFail and then call sendSuccess (an error will be thrown).
• Functions passed to Task can optionally create a cancel (like above) otherwise cancel will be an no-op.
• It's lazy! The function passed on Task creation is only called when run is invoked.
• There is no error catching in this Task implementation. Errors are not thrown or caught from within a Task. There are failure values but these are not the same thing as errors -- think of them as "bad news".

Chaining

function getUser(id) {
  return new Task((sendFail, sendSuccess) => {
    // AJAX request to get a user with id
    sendSuccess({ user });
  });
}
 
function getFollowers(username) {
  return new Task((sendFail, sendSuccess) => {
    // AJAX request using username
    success([followers]);
  });
}
 
getUser()
.map(user => user.name)
.chain(getFollowers)
.run(
  fail => {},
  success => {
    // success === [followers] (if all went well)
  }
});

Memoization

A promise-like feature that allows you to hang on to values already processed within a Task. Computations don't get re-run.

var count = 0;
const task = new Task((sendFail, sendSuccess) => {
  const id = setTimeout(() => {
    sendSuccess(++count);
  }, 1000);
  return () => { clearTimeout(id); };
});
 
const taskOnce = task.memoize();
 
taskOnce.run(
    fail => {},
    success => {
      // success === 1
      // count === 1
    }
});
taskOnce.run(
    fail => {},
    success => {
      // success === 1
      // count === 1
    }
});

Parallel Tasks

Parallelize multiple Tasks. Returns an array of successs. If one of the Tasks sends a fail then the cancelation functions for all other Tasks (not yet completed) will be called.

var count = 0;
function createTask(to) {
  var order = ++count;
  return new Task(sendFail, sendSuccess => {
    var id = setTimeout(() => {
      sendSuccess(order);
    }, to);
    return () => {
      clearTimeout(id);
    };
  });
}
 
Task.all([
  createTask(100),
  createTask(500),
  createTask(0)
]).run(
  fail => {},
  success => {
    // count === 3
    // success === [3, 1, 2]
  }
});

Specifications compatibility

Task is compatible with Fantasy Land and Static Land implementing:

• Semigroup
• Monoid
• Functor
• Bifunctor
• Apply
• Applicative
• Chain
• Monad

How is this Task different than Data.Task or Fun-Task

• This Task throw an error if you attempt to call sendFail or sendSuccess if either has already been called.
• This Task offers a memoization method that allow you to treat Tasks more like promises so computations don't get called more than once if multiple parts of your code call run on an Task.
• No special or magical error catching involved.

Credits

A lot of code was inspired and stolen directly from data.task (Quildreen Motta) and fun-task (Roman Pominov).