Use Less. Do More. JavaScript on steroids.

Use Less. Do More.

A cross-platform JavaScript toolbox for writing complex web applications. Currently in pre-release stage, missing some basic documentation and being under heavy development.

Installing | Building | Wiki

> npm install useless

Client side tools that depend on reflection utility (e.g. test framework / stack traces UI) work only in WebKit-based browsers at the moment. Server side tools are built on top of Node.js technology. The rest of the code base should be cross-browser (at least it attemps to).

Database-related utility depends on MongoDB (not included in npm dependency list, should install manually). Anyway, it does not work well at the moment...

> node example.js

If everything's ok, example app will be running at http://localhost:1333. Currently there's not much example code, but it's on the way.

  • ./base platform-independent part
  • ./client browser-related utility
  • ./server app framework for Node.js
  • ./build built source (full/stripped/minified)
  • ./useless.js header file for the build tool (contains include directives)
  • ./useless.micro.js header file (an example of reduced build config)
  • ./example.js example application
  • ./example static content for example.js app

It started a year ago as a pet library for my freelance projects, but recently it has kinda grown out of control, so I decided to make it public domain.

Its main (unreleased for now) feature will be a flexible schema layer over MongoDB which allows to configure various data access widgets (table views, popover selections, form dialogs) directly from that schema, providing real-time synchronization with remote database. It also will act as declarative contract for API validation / security layer. Thats why there's so many code related to abstract data processing / type matching.

It also features various widgets with rendering virtualization, allowing to render vast amounts of data with smooth scrolling and no pagination (TableView, TileView, ListView). They're already there, but not yet working as they depend on that schema mentioned before. Need to separate this thing from my elder private projects properly... Currently I'm focused to make the basic bootstrap code work well, as this thing is now used as a front-end library in a couple of large scale projects run by a company I'm employed. So at least the base part should be production quality soon.

Stay tuned and thanks for your attention!

Feel free to leave feedback / submit pull requests if you find any of these things helpful. Documentation is pending, check Wiki for updates.

./base features

How-to & Examples

  • $prototype / $extends
  • Smart property declarations
  • $static methods / properties
  • $memoized properties
  • Tag groups on members ($static: { ... })
  • RTTI
  • Pluggable macros for custom syntax
  • Member aliases
  • $final
  • $traits / a.k.a. mixins / combinatoric-style alternative to inheritance
  • $aspect / Aspect Oriented Programming / declarative method binding
  • $singleton
  • Reflection (can read prototype name and file, via Prototype.$meta)
Vec2 = $prototype ({
    /*  Static property:
    zero: $static ($property (function () { return new Vec2 (0, 0) })),
    /*  Static method: (a, b)
    dot: $static (function (ab) { return a.x * b.x + a.y * b.y }),
    /*  Tag groups for convenience
    $static: {
        unit: $property (function () { return new Vec2 (1, 1) }),
        one:  $alias ('unit') }, // member aliases 
    /*  Constructor
    constructorfunction (xy) { this.x = x; this.y = y },
    /*  Instance property (.length)
    length: $property (function () { return Math.sqrt (this.x * this.x + this.y * this.y) }),
    /*  Instance method
    addfunction (other) { return new Vec2 (this.x + other.x, this.y + other.y) } })
/*  Inheritance (relies on native JavaScript prototype semantics)
BetterVec2 = $extends (Vec2, { /* ... */ })

How-to & Examples

  • Binds own methods to this automatically
  • Manages bindable $trigger / $barrier / $observableProperty members
  • Tracks bound components / auto-unbinds upon deinitialization
  • Holds parent-child relationship / handles automatic deinitialization
  • Enables $traits to chain into method calls by overlapping method definitions
  • Enforces configuration contracts ($requires, $defaults)


  • Extends underscore.js _ namespace
  • Functional primitives busybox (predicates, operators / higher order stuff)
  • Binding to tail of argument list / flipping argument list
  • Limiting number of arguments (arity)
  • Infix interface (as Array/String/Function extensions)
  • Handy string-processing operators
  > [[1,2], [3,4]].zip (_.seq (_.sum, _.appends ('_foo'), _.quotesWith ('()')))
  < ["(4_foo)", "(6_foo)"]


Datatype-abstract (works over arrays/objects/scalars):

  > _.map2 ({ one: '1', two: '2' }, _.prepends ('foo_').then (_.appends ('_bar')))
  < { one: "foo_1_bar", two: "foo_2_bar" }

Structure-abstract ('sees through' structure of arbitrary complexity):

   > _.mapMap ({ foo: { bar: 1, baz: [2, 3] } }, (10))
   < { foo: { bar: 10, baz: [12, 13] } }


Continuation-passing style ( versions of underscore.js primitives:

   function searchRemoteFilesForText (textthen) {
      _.cps.find (['file1.txt', 'file2.txt', 'file3.txt'], function (namereturn_) {
          $.get (name, function (fileText) {
              return_ (fileText.indexOf (text) >= 0) }) }, then) }
   log ( ([1,2,3],     _.constant ('stub'))      // prints ['stub','stub','stub'] ([1,2,3], _.cps.constant ('stub'), log) // prints ['stub','stub','stub'] 
   cachedReadFile = _.cps.memoize (_.tails ($.get, 'text'))
   cachedReadFile ('/useless.js', log) // prints contents of /useless.js 

Sequential composition of asynchronous operations:

  _.cps.sequence (doRoutine, waitUntilAssertionsComplete, done) ()

Task pooling (parallel map/reduce with limit on maximum concurrently running tasks):

  _.mapReduce (array, {
                  maxConcurrency: 10,
                  nextfunction (itemindexnextskipmemo) { ... },
                  completefunction (memo) { ... })

$interlocked (puts a function under concurrency lock)

  readFilesSequentially = $interlocked (function (releaseLockfiledone) {
                                         $.get (file, done.then (releaseLock), 'text') })
  readFilesSequentially ('file1.txt', log)
  readFilesSequentially ('file2.txt', log) // waits until file1.txt is read 


  • _.trigger, _.triggerOnce / one-to-many broadcast
  • _.barrier / synchronization primitive
  • _.observable / state change notifications

Raw API (same for every mentioned primitive):

var mouseMoved = _.trigger ()
/*  Binding
mouseMoved (function (xy) { }) // bind 
mouseMoved (someCallback)        // bind another 
/*  Calling
mouseMove (12, 33)               // call 
/*  Unbinding
 */ (someCallback)     // unbinds specific listener ()                 // unbinds everything (someCallback)             // unbinds callback from everything it's bound to 

Using $component:

Compo = $component ({
    didLayout:     $trigger (),
    layoutReady:   $barrier (),             // it's like document.ready 
    value:         $observableProperty (),  // for property change notifications 
    initfunction () {
        doSomeUselessAsyncJob (function () {
           this.layoutReady () }) }, // signals that layout is ready 
    doLayoutfunction () {
        this.didLayout () } })       // simply call to perform multicast 
compo = new Compo ()
compo.didLayout (function () {
    /*  Gets called whenether layout has rebuilt */ })
compo.layoutReady (function () {
    /*  Postpones until DOM is ready.
        If already, calls immediately (like document.ready) */ })
compo.valueChange (function (valueoldValue) {
    /*  Gets called whenether property has assigned distinct value */ })
compo.value = 10 // simply assign a value to notify listeners 
compo.value = 10 // won't trigger, as not changed 

Raw API:

_.onAfter   (Player.prototype, 'move', function (xy) { /* this will execute after move calls */ })
_.onBefore  (Player.prototype, 'move', function (xy) { /* this will execute before */ })
_.intercept (Player.prototype, 'move', function (xyoriginalMethod) { (this, x, y) })

Using $component + 'once' semantics:

Button = $component ({
    layout: $bindable (function () { /* ... */ }) })
button = new Button ()
button.layout.onceBefore (function () { log ("I'm called before next layout()") })
button.layout ()
button.layout () // won't print anything 

Using $aspect:

AddsLoggingToButton = $aspect (Button, {
    beforeCreatefunction () { ('Button is about to be created') },
    afterDestroyfunction () {   ('Button is now destroyed') } })


Working with ranges:

    _.lerp  (t, min, max)  // linear interpolation between min and max 
    _.clamp (n, min, max)  // clips if out of range 
    /*  Projects from one range to another (super useful in widgets implementation)
    _.rescale (t, [fromMin, fromMax], [toMin, toMax], { clamp: true })

Vector math (Vec2, Transform, BBox, Bezier, intersections):

   var offsetVec = this.anchor.sub ( (
                       Bezier.cubic1D (
                  (direction.normal, upVector), 0, 1.22, 0, 1.9))
   var where = this.bodyBBox.nearestPointTo (this.anchor, this.borderRadius)
   domElement.css (BBox.fromPoints (pts).grow (20).offset (position.inverse).css)

  • Cross-platform handling of uncaught exceptions
  • Uncaught exceptions pass through network API calls
  • Client displays server's exceptions as if it was single environment
  • Callstack API for access at arbitrary location (for reflection purposes)
  • Strips third party calls (clean mode)
  • Fetches source code (local/remote)
  • Nice output
    • Console mode (replaces default Node.js exception printer)
    • GUI mode (a pop-up dialog with expandable source lines, ./client/Panic.js feature)

How-to & Examples

  • Tests before code
  • Tests as documentantion
  • Rich library of assertions
  • Asynchronous / nested assertions
  • Intercepts global log, displaying it in running assertion context
  • Custom assertions
  • Humane error reporting

Reference / examples

/*  A cross-platform alias for the object representing global namespace.
    Use instead of `window` for cross-platform code.
$ = 5 
/*  Implements `alert` for NodeJS platform.
alert ('foo')
/*  Working with uncaught exceptions
_.withUncaughtExceptionHandler (function (e) { throw e /* re-throw */ }, // adds handler to chain 
                                function (done) {
                                    done () })                           // removes handler from chain 
Platform = $singleton ({ $property: {
    engine: ... // 'browser' / 'node' 
    system: ... // 'iOS' / 'Android' / undefined for everything else 
    device: ... // 'iPad' / 'iPhone' / undefined for everything else 
    touch:  ... // true for touch-enabled devices 
    Browser: ... // true in browser 
    NodeJS:  ... // true in Node 
    iPad:    ... // true on iPad, 
    iPhone:  ... // true on iPhone, 
    iOS:     ... // true on any iOS device } }) 

./client features

A pack of handy jQuery extensions. Biggest thing here is drag & drop utility ($.fn.drag), which is utilized by countless number of widgets I made. It also compatible with mobile devices (iOS / Android).

$(handle).drag ({
    startfunction ()             { return this.leftTop () },
    move:  function (memooffset) { this.css (memo.add (offset).asLeftTop) } })

Included in useless.devtools.js distribution

Ever struggled with bugs in JavaScript? Then Panic.js is your instant best friend. Delivers better error diagnostics to Chrome, Safari and Firefox.

Live demo _(clickable)_:

  • Cross-platform uncaught exception handling (works around incomplete 'onerror' impl. in Safari).
  • Maintains callstack persistence across async call boundaries (addEventListener, setTimeout).
  • Loads and displays expandable source lines for fast identification.
  • Hides third party code by default (can be displayed by pressing 'more').
  • Grouping of message duplicates.
  • Grouping of same source line repeated consequently.
  • Complete API for it's internals (exception handling / callstack access) — TBD

Yes, and this is "debug tools" too, but more specific. It is not a replacement to WebInspector, but a power-up. WebInspector is just not good enough for the quick identification of typical errors that occur often during development process. In most cases, they require no detailed inspection of the full source code — you can identify problem cause just by briefly looking at the source line. Displaying full info, as WebInspector does, would slow down things: you need to scroll through tons of text, expand cryptic traces by clicking, and then clicking and waiting again to load sources in separate window.

And because WebInspector is a separate tool, you might not even know that something's broken, until its opened. You can imagine Panic.js as a small and fast subset of WebInspector, inlined into your page until it goes to the production, like a scaffolding.

Simply link the script to a page, and it will configure itself automagically™. Requires underscore and jQuery.

Before use, consider that the distribution contains full Useless toolkit, where the diagnostics-related utility [responsible for all the magic behind Panic.js] appear a small fraction of entire codebase. In other words, it brings a holy shitload of code, which can possibly cause all sort of compatibility-related issues.

But in most cases, it should work out of the box. And supposing that you don't ever plan using Panic.js on production server (why would you), size of the script doesn't matter that much.

P.S.: it does not display source lines if executed from local HTML file (as sources are read by XHR requests).

Included in useless.devtools.js distribution

A nice console.log alternative for those who like log-driven debugging. Based on log.js (acts as a write backend for it) — which is ultimately based on stringify.js and reflection.js superpowers (check it out).

Renders incoming log calls into semi-transparent overlay at the bottom of the viewport. It is transparent for mouse input (so it doesnt mess with page interaction) and can be toggled by ~ key (hello Quake). Much nicier than default console.log in many cases.

LogOverlay.init ()
log.('Hello world...')
log.('...and hasta la vista, baby')

XHR requests to communicate with ./server part.

  • DataManager.js
  • Collection/Collection.js
  • Collection/RemoteCollection.js
  • Collection/FilterChain.js

Bits and pieces of real-time database synchronization engine, taken from some previous project. Needs documentation and examples. Rather unusable for now.

./server features


require ('useless')
UselessApp = $component ({
    apifunction () { return {
        '/':           this.file ('./static/index.html'),
        'hello-world': this.helloWorld } },
    helloWorldfunction (context) { context.success ('Hello world!') },
    $traits: [
        require ('useless/server/exceptions'),
        require ('useless/server/tests'),
        require ('useless/server/deploy'),
        require ('useless/server/api'),
        require ('useless/server/io'),
        require ('useless/server/http') ],
    initfunction (then) {
        then ()
        log.ok ('App started') } })
module.exports = { initfunction () { return new UselessApp () } }

Following are $traits defined at useless/server:

  • api.js URL mapping
  • appcache.js AppCache back-end
  • auth.js authorization management (boilerplate, for demo purposes)
  • deploy.js self-deployment protocol (automatic builds)
  • devtools.js APIs for Git / source code access (developer mode)
  • entity.js journaled CRUD for collections (high level DB access)
  • exceptions.js error handling for requests
  • history.js journal for DB operation
  • http.js request serving basics
  • io.js basic I/O for requests
  • templating.js basic templating (via underscore)
  • tests.js self-test on startup, adds $test/$tests syntax for server $traits + new asserts
  • uploads.js file uploads
  • uptime.js uptime tracking / restart()
  • websocket.js WebSocket utility (peer tracking / auth / multicast)


Type npm install useless in root directory of your project.

  1. Go to node_modules subfolder of your project
  2. Run git clone, go to useless folder
  3. Run npm install to install dependencies
  4. Optionally, run node build.js to test if everything's ok

Go to build folder and pick useless.js. For minified version (with unit tests stripped) pick useless.min.js. This one is ready to be used in production setup.

This version includes only the platform-independent part of the library, not including any of the ./client features. If you need any of these, you can either link them them in browser code separately, or make custom build file with the additional files included (see instructions below).


Build command:

node build.js <header-file-1> <header-file-2> ... <header-file-N> <output-folder> [no-compress]

For generic build, run node build.js ./useless.js ./build, or simply

> node build.js

It will generate ./build/useless.js by substituting $include directives found in header file. Produced result will undergo stripping of tests and comments, and then finally compiled using Google Closure Compiler, outputting minified result to ./build/useless.min.js

To make reduced/extended distribution (with some submodules disabled or enabled), you can create your own version of default header file, commenting out unneeded $include directives or including additional ones.

There exists ./useless.micro.js as an example of reduced build. Running node build.js ./useless.micro.js ./build will produce ./build/useless.micro.min.js as output.

You can run build.js under nodemon (which can be installed from npm). This will trigger automatic re-builds on source change.

nodemon build.js <header-file> <output-folder>

This will work for applications that dont rely on useless/server to implement app lifecycle. For frequent re-builds, you may turn off compression, re-building only useless.js (Google Closure Compiler has limited call quota per IP):

> nodemon build.js no-compress

Applications that are based on top of useless/server can easily enable automatic builds feature by adding following $traits to main application component:

$traits: [        
        require ('useless/server/exceptions'),
        require ('useless/server/tests'),
        require ('useless/server/deploy'),

This will add test & build phase to app startup sequence, aborting if something went wrong.

For re-scheduling startup on source change, run your application under nodemon or supervisor. Important notice: you should add ./node_modules/useless/build/ folder to .nodemonignore file in root directory of your project, to prevent restart loop.

Currenly it re-builds only useless.js, with no compression applied.