The g of a mack and the steez of a geezer.
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


The g of a mack and the steez of a geezer.


Steez helps you create well behaved streams in a very unobtrusive manner. It does this by hijacking your outgoing data events, implementing buffering, providing accurate pause()/resume() behaviour and trying really hard to play nicely with all manners of pipe() usage.


Implementing pause()/resume()/end()/etc semantics every time you write some code that you want to use as a stream is annoying. Steez tries to make this a thing of the past by providing an extendable base class with some transparent magic built in to make your stream behave nicely.


  • Transparent data event queueing (by hijacking emit()) (more)
  • Accurate pause(), resume(), end(), destroySoon() and destroy() implementations out of the box
  • Ability to override certain behaviours on a per-implementation basis (more)
  • Multi-target stream synchronisation (more on this below)


Steez internally hijacks your emit() method, catching data events and queueing them if the stream is in a paused state (if this.paused is true). When emitting data, it's advised that you check for the state of this.paused; if it's not 0, hold off on sending any more data until Steez emits a drain event.

WARNING: If you continue to emit data at a high rate while the stream is paused, there's a chance you'll blow out your memory usage as events will be indefinitely buffered.

Overridable Methods

Steez has a few methods you can implement to alter its behaviour; they are as follows:

can_destroy: return a truthy value if the stream has no pending tasks and can be destroyed safely. This is useful if you perform any asynchronous tasks in your stream, as they will not count towards the stream's internal queue.

on_pause: this is called (if it exists) when Steez goes from being not paused to paused. Note that this is not as simple as just being called every time pause() is called, due to the multi-target stream synchronisation.

on_resume: this is called (if it exists) when Steez goes from being paused to not paused. This is the same as on_pause in that it's not just called whenever resume() is called, but rather when all streams have resume()ed the stream.

Multi-Target pipe()

With a regular stream, if you pipe to more than one target and more than one of those targets initiates or relays pause() or resume() calls, you can end up with some very strange behaviour.

Take this code for example:

var fs = require("fs"),
    util = require("util");

var src = fs.createReadStream("/mnt/my_ssd/src"),
    fast_dst = fs.createWriteStream("/mnt/my_ssd/dst"),
    slow_dst = fs.createWriteStream("/mnt/my_floppy_drive/dst");


In this case, you would logically expect the src stream to be limited to the slower of the two dst streams. This, however, is not the case. In reality, the you will end up with the src stream flipping rapidly between paused and unpaused. This is why:

  1. slow_dst stream will pause() the src stream as it will block on writing much quicker than fast_dst.
  2. Before slow_dst has become ready for writing again, fast_dst will resume() src as its write buffer will have emptied (after slow_dst called pause() on src).
  3. Upon emitting another chunk of data, slow_dst will immediately pause() src again, as it's still not ready for writing.
  4. Points 2 and 3 will be repeated for a while, until either fast_dst finishes or slow_dst collapses under the weight of its own internal buffering mechanism (hey, it could happen).

To solve this, Steez keeps a record of which pipe()ed streams have asked to pause() or resume() the Steez instance and only resumes emitting data after all targets are ready.


Available via npm:

$ npm install steez

Or via git:

$ git clone git:// node_modules/steez


#!/usr/bin/env node

var fs = require("fs"),
    util = require("util");

var Steez = require("steez");

function ChunkReverser() {;
util.inherits(ChunkReverser, Steez);

ChunkReverser.prototype.write = function write(data) {
  this.emit("data", Buffer(data.toString().split("").reverse().join("")));

  return !this.paused && this.writable;

var chunk_reverser = new ChunkReverser();

var a = fs.createReadStream("./"),
    b = fs.createWriteStream("./data.out");



3-clause BSD. A copy is included with the source.