JavaScript needs modules | Hacker News Reader

JavaScript needs modules

blog.mozilla.com

47 pointssant0sk137 comments

Direct link to pdf: http://www.ccs.neu.edu/home/dherman/talks/javascript-needs-m...

Seems like CommonJS but with keywrords added to the language instead of magic functions/free variables. As such it's six of one, half dozen the other as far as I can tell.

From a cursory glance, this seems like mostly useless syntactic sugar. I don't exactly see what is the functional different between

  module A {
    export var myPublic = 1
    var myPrivate = 2
  }

and

  var A = new function() {
    this.myPublic = 1
    var myPrivate = 2
  }

except that the first will clash with a bunch of existing script that use "module" and "export" as variable names. And the idea of making "load" a keyword in page 26 is just plain silly.

You missed two bits:

  - the goal is to not pollute the global namespace as we do today
  - the module/export/load are a part of an opt-in dialect of javascript when you specify "text/es-harmony" as the language.

>> the goal is to not pollute the global namespace

Right, like this, I guess

  new function() {
    var jQuery = ...
  }

I think this document is either omitting the important details or it simply doesn't address the namespacing problem correctly. Remember when Prototype.js had a document.getElementsByClassName? Fun times when that broke.

>> the module/export/load are a part of an opt-in dialect

That doesn't seem very well thought-out. How would one access a module declared in a <script type="text/es-harmony"> from a <script type="text/javascript">? It's a pretty realistic scenario (e.g. say you have a non es-harmony codebase and you want to add a es-harmony library)

>> I think this document is either omitting the important details or it simply doesn't address the namespacing problem correctly. Remember when Prototype.js had a document.getElementsByClassName? Fun times when that broke.

Interfering with other modules' exports is not possible. But modifying the properties of arbitrary objects is just as possible as ever. Modules are all about fixing scope, not about locking down objects.

To put it differently, the document is not a module, it's an object. The module system has nothing to do with it.

>> How would one access a module declared in a <script type="text/es-harmony"> from a <script type="text/javascript">? It's a pretty realistic scenario (e.g. say you have a non es-harmony codebase and you want to add a es-harmony library)

This is accounted for in the design. The global object is not in the scope chain of Harmony code, but it is still available to Harmony code via a standard binding in the standard library, and the Harmony modules are made available to legacy JS as module instance objects in the global object. So communication is available in both directions.

>> Harmony modules are made available to legacy JS as module instance objects in the global object.

Does this mean a module is visible like this?

  window.jQuery
  window.Prototype

That's kinda what already exists... For that matter, any variation of that (e.g. window.modules.jQuery) can be done with the patterns you mentioned at the beginning of the document.

>> The global object is not in the scope chain of Harmony code

Ah, I see. That's certainly a new feature, but what's a use case for that? I find it hard to justify giving up a few fairly common variable names just for knowing that window.i or whatever is not polluted. It kinda feels like just shifting the name collisions around.

the goal is to not pollute the global namespace as we do today

What is the other option? The jQuery in "import jQuery.$" has to refer to something. Does he want to create a "module namespace" with special "module objects" not referenceable from normal variable scope? No thanks: that sounds like Java, not JavaScript.

the module/export/load are a part of an opt-in dialect of javascript when you specify "text/es-harmony" as the language

Why does Mozilla need to build dialects into their browser for something which can be done with a simple compiler? Between "create a little module syntax compiler which works in all browsers now" and "wait for all of the borwsers to implement the text/es-harmony dialect", I'd rather see the former.

Because the Mozilla developers are the same wise guys who think IndexedDB is superior to SQLite!

If they keep this up, web developers will eventually abandon Firefox. Firefox is starting to be the new IE6.

JavaScript 1.7 already supports a very similar module pattern. For example:

    myModule: {
      // don't do anything if myModule is already included
      if (myModule) break myModule;
      
      let this_is_private = ...;
      var myModule = ...;
      // myModule is public and won't pollute the
      // global NS if it isn't included globally.
    }

Joose has a Modules facility that works pretty nicely both in the browser and in other contexts, e.g. Node.js. When combined with the "use" facility of JooseX.Namespace.Depended, I think you get a similar functionality as to what's being described:

http://openjsan.org/doc/s/sa/samuraijack/Joose/3.010/lib/Joo...

http://openjsan.org/doc/s/sa/samuraijack/JooseX/Namespace/De...

I think until there is some native namespace/module solution for JS, fusebox [http://github.com/jdalton/fusebox#readme] will fill an excellent gap (upon release, soon!!!) for this sort of need.

I am really excited about the future of Fuse and fusebox in web development. John David-Dalton's work on both is excellent, and I am happy I have been apart of the exciting development process.

Please dont break compatibility when it is possible to avoid it.

js is coming to the server quite nicely, most modern browsers support it well too... Even microsoft starts to take the web seriously with IE9.

I am not convinced mozilla fondation is in its role breaking compatibilities when avoidable

You can get modules right now, in every browser, with CoffeeScript. It compiles nicely to standard JS:

http://coffeescript.org

I'm a big fan of CS, but I wasn't aware it has module support. Are you talking about the class system?

Sorry, I should've been clearer: I meant that CoffeeScript gives you the safety benefits of modules, just by breaking your code into separate .coffee files (unless you use '--no-wrap'), since variables aren't shared unless explicitly exported.

CoffeeScript does many things at the language level, but modules is not one of them.

Maybe in 0.8 or 0.9, Jeremy? ;-) It could "solve" the "problem" of wrapping compiled code in anonymous functions and having to attach exports to `window`.

Ah, but all of your compiled code is already wrapped in a safety function. (perhaps that's what Trevor was referring to). As for attaching to exports or window, that's going to have to be something that you do by hand regardless. If everything is added to the exports, then there's no point in wrapping in the first place...

Well, what I'd like to do is have e.g.

mymodule.coffee:

    x: 5
    y: (z) -> z * z

main.coffee:

    load("mymodule")
    puts mymodule.y(5)

I could set this up manually, of course, but if I wanted to do things manually I wouldn't be using CoffeeScript. :p

Yes, I meant that the JavaScript "module pattern" is implemented by default, and the language prevents you from declaring globals by mistake. So, you get most of the benefits of the proposed JS module syntax.

It's a shame that somebody point at this obvious feature on 2010. Javascript lacks a lot of features from scripting languages.

Javascript is a scripting language -- and off the top of my head, I can't think of any scripting language with modules (maybe VBScript?)

Lua has packages.

http://www.lua.org/pil/15.html

Python, Perl, Ruby, Lua, R, Matlab, etc...I can think of more that do than don't.

None of those are scripting languages, except maybe Matlab -- I haven't used it enough to say.

Just because a language is dynamically typed doesn't make it a "scripting language".

Dynamically typed, no. Interpreted, yes.

JavaScript is typically implemented with a JIT compiler, just like Java or C#. Python, Ruby, and Lua use virtual machines (like Java used to be).

That's why I've always maintained that Java is a scripting language ;)

typically? JIT for javascript is only a semi-recent innovation. I'd say most javascript runtimes pre-2009 didn't have a JIT.

> (like Java used to be)

So Java doesn't run on a virtual machine anymore?

What? Lua isn't a scripting language? Can you flesh that out a bit?

(So is TCL, but I don't think it has a module system.)

Lua is a full programming language -- you can use it to write a complete application without depending on external third-party tools.

I usually take "scripting language" to mean "a language only suitable for combining pre-written components", since that seems to be the only useful definition. Other definitions are too broad (containing such languages as Python, Java, Haskell, and Forth) or too narrow (excluding QuakeScript, JavaScript, or Bash).

In particular, I refuse to call any language a "scripting language" merely because it uses dynamic typing, implicit compilation, or high-level concepts.

I think your definition is wrong, just go with wikipedia's definition: http://en.wikipedia.org/wiki/Scripting_language

Javascript can be run at the command-line just like ruby, python, and the rest. You can write a full application in pure javascript without depending on external third party tools too -- see stuff like nodejs.

Also, I don't know of anyone that would consider Haskell to be a scripting language. If we consider Haskell a scripting language, we might as well consider Java, C, C++ to be scripting languages too...

According to the Wikipedia definition, Haskell is a scripting language (it can be dynamically typed, interpreted, embedded, etc). That's why I don't think it's a useful definition.

How do you write a full application in JavaScript? The language doesn't even define any way to open a file. NodeJS is a third-party tool -- it's not part of the JavaScript language.

While I agree that "scripting language" is a bit loosely defined (though nothing like "object-oriented"), Lua was written with a "hard and soft layers" (http://c2.com/cgi/wiki?AlternateHardAndSoftLayers) approach in mind. To me, that seems like the defining characteristic of a scripting language.

I'm not saying Lua isn't a full programming language - it's actually my language of choice for day-to-day hacking, supplemented with C as necessary. It was clearly designed to accommodate projects where it isn't primary language, though.

I think it is better to replace JavaScript with a more powerful language like C

It's all about using the right tool for the job.

Reply on news.ycombinator.com