Serializing a stream

gistfile1.js

// Sometimes we need to handle stream results serially...
// The question is, is there a better / more obvious way to do this?

var fs = require('fs'),
  streamer = require('streamer');

function serialize(source) {
  var queue = [];
  
  return function stream(next, stop) {
    var processing = false,
      completed = false;
    
    source(function sourceNext(x) {
      if (processing) {
        queue.push(x);
        return;
      }
      
      processing = true;
      next(x, function() {
        processing = false;
        if (completed && !queue.length) {
          stop();
        } else {
          sourceNext(queue.shift());
        }
      });
    }, function(err) {
      if (err) return stop(err);
      completed = true;
    });
  };
}

// a simple async list implementation
function asyncList(arr) {
  return function stream(next, stop) {
  	var l = arr.length;
  	
  	arr.forEach(function(v, k) {
	    setTimeout(function() {
	      next(v);
	      if (k===l-1) stop();
	    }, 50);
  	});
  };
}

serialize(asyncList([0,1,2,3]))(function(x, next){
  console.log('wait: ',x);
  setTimeout(function() {
    console.log('continue: ',x);
    next();
  }, 50);
}, function() {
  console.log('completed!');
});

@Gozala Ahh, interesting. Thanks for the reply; I'll have to take a look.

Have you seen node-lazy? It's an event emitter / node.js Stream based approach to lazy streams. It's more familiar, but being non-functional, it would seem it's not "trivially parallizeable" as MapReduce / no-side-effect functional solution is. Also though, aside from the issue in the gist, IMHO streamer feels more elegant.

@CrabDude To be fully honest, back when I started with streamer I did not knew there was anything similar. Later people pointed me out node-lazy and streamjs which I looked at, but quickly rejected mainly because of method based approach instead of functional. At this point I'm no longer sure that function based approach was indeed better as I experiment with prototype based solution myself. Once I'll get more confidence with best possible solution, I'm planning to talking with both @pkrumins and @dionyziz about merging our efforts.