walk-machine

Control callback flow through a finite state machine with functions that do just one thing each.

Installation

npm install walk-machine

Usage

var walkMachine = require('walk-machine');

var stateMap = {
  start: {
    work: db.get,
    params: [
      {
        location: '/somewhere/entries',
        ignoreTemporaryEntries: true
      }
    ],
    checkError: checkLoadError,
    next: getStateAfterLoadEntries
  },
  loadImages: {
    work: addImagesToEntries,
    next: 'renderEntries'
  },
  renderEntries: {
    work: callRender
  }
};
walkMachine(stateMap, handleError);

Why

Let's say you have a few async tasks to coordinate:

• Load some "entries". Getting them from the db may result in a NotFoundError, but we want to ignore that, while bailing for all other errors.
• If those entries all have images, then render the entries.
• If any of the entries are missing images, find images for them and add them to the entries. Then, render the entries.

This is what it'd look like if implemented with async.waterfall:

async.waterfall(
  [
    start,
    branch,
    callRender
  ],
  handleError
);

function start(done) {
  var getOpts = {
    location: location,
    ignoreTemporaryEntries: ignoreTemporaryEntries
  };
  db.get(getOpts, checkEntries);

  function checkEntries(error, entries) {
    if (error.type === 'NotFoundError') {
      // We can ignore this.
      done(null, entries);
    }
    else {
      // Yes, it's a real error.
      done(error);
    }
  }
}

function branch(entries, done) {
  if (entries.every((entry) => entry.image)) {
    callNextTick(null, entries);
  }
  else {
    addImages(entries, done);
  }
}

function callRender(entries, done) {
  render({
    entries: entries,
    parentElement: document.getElementById('root')
  });
  callNextTick(done);
}

function addImages(entries, done) {
  var q = queue(1);
  entries.forEach((entry) => q.defer(addImage, entry));
  q.awaitAll(done);
}

function addImage(entry, addDone) {
  async.waterfall(
    [
      findImageForEntry, // Async function defined externally.
      addImageToEntry
    ],
    addDone
  );

  function addImageToEntry(image, entry, addToEntryDone) {
    entry.image = image;
    callNextTick(addToEntryDone, null, entry);
  }
}

There's a few problems here:

• The call to waterfall implies that the control flow is all right there, but some of it is also out in branch and there's a sub-waterfall over in addImage.
• start has to do some awkward error-checking. Normally, we let waterfall handle the error checking, but we don't want NotFound errors to interrupt the flow. So, task functions handle errors in two different ways.

Waterfall's best for non-branching task sequences.

This is what this program looks like implemented with promises:

var startPromise = new Promise(start);
startPromise.catch(handleError);
startPromise.then(branch);

function start(resolve, reject) {
  var getOpts = {
    location: location,
    ignoreTemporaryEntries: ignoreTemporaryEntries
  };
  db.get(getOpts, checkEntries);

  function checkEntries(error, entries) {
    if (error.type === 'NotFoundError') {
      // We can ignore this.
      resolve(entries);
    }
    else {
      // Yes, it's a real error.
      reject(error);
    }
  }
}

function branch(entries) {
  if (entries.every((entry) => entry.image)) {
    return Promise.resolve(entries)
      .then(callRender);
  }
  else {
    return Promise.all(entries.map(makeAddImagePromise))
      .then(callRender);
  }

  function makeAddImagePromise(entry) {
    return Promise.resolve(entry)
      .then(findImageForEntry)
      .then(addImageToEntry);
    }
  }
}

function addImageToEntry({image, entry}) {
  entry.image = image;
  return entry;
}

function callRender(entries) {
  render({
    entries: entries,
    parentElement: document.getElementById('root')
  });
}

OK, there's a lot of problems there, too. TODO: Check promises code.

Here's how we can do it with a state machine:

var walkMachine = require('walk-machine');

var stateMap = {
  start: {
    work: db.get,
    params: [
      {
        location: '/somewhere/entries',
        ignoreTemporaryEntries: true
      }
    ],
    checkError: checkLoadError,
    next: getStateAfterLoadEntries
  },
  loadImages: {
    work: addImagesToEntries,
    next: 'renderEntries'
  },
  renderEntries: {
    work: callRender
  }
};
walkMachine(stateMap, handleError);

function handleError(error) {
  if (error) {
    console.log(error, error.stack);
  }
}

function checkLoadError(error) {
  if (error.type === 'NotFoundError') {
    // No, we can ignore this.
    done();
  }
  else {
    // Yes, it's a real error.
    done(error);
  }
}

function getStateAfterLoadEntries(error, entries, done) {
  if (entries.every((entry) => entry.image)) {
    done(null, 'renderEntries');
  }
  else {
    done(null, 'loadImages');
  }
}

function addImagesToEntries(entries, done) {
  var q = queue();
  ...
  q.awaitAll(done);
}

function callRender(entries, done) {
  render({
    entries: entries,
    parentElement: document.getElementById('root')
  });
  callNextTick(done);
}

That's still a lot of code! But:

• All of the states are explicitly listed up front.

• Each function either does "work" or it figures out the next state. None of them do both.

• There's fewer "prep" functions – functions that do nothing but gather together parameters to pass to another function. callRender does that, but at least we are able to set db.get up using walkMachine instead of having to create a loadEntries function that looks like this:

  function loadEntries(location, ignoreTemporaryEntries, done) { db.get({location: location, ignoreTemporaryEntries: ignoreTemporaryEntries}, done); }

Tests

Run tests with make test.

License

The MIT License (MIT)

Copyright (c) 2016 Jim Kang

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.