user defined object equality for a set in harmony (es6) - javascript

I have a problem where I'm generating many values and need to make sure I only work with unique ones.
Since I'm using node js, with the --harmony flag, and have access to harmony collections, I decided that a Set may be an option.
What I'm looking for is something similar to the following example:
'use strict';
function Piece(x,y){
this.x = x
this.y = y
}
function Board(width,height,pieces){
this.width = width
this.height = height
this.pieces = pieces
}
function generatePieces(){
return [
new Piece(0,0),
new Piece(1,1)
]
}
//boardA and boardB are two different but equivalent boards
var boardA = new Board(10,10,generatePieces())
var boardB = new Board(10,10,generatePieces())
var boards = new Set()
boards.add(boardA)
boards.has(boardB) //return true
Now normally to achieve this in another language, say c#, I would expect to have to implement an equals function, as well as a hash code generating function for both Board and Piece. Since I'd expect the default object equality to be based on references.
Or perhaps use a special immutable value type (say, a case class in scala)
Is there a means to define equality for my objects to solve my problem?

Is there a means to define equality for my objects to solve my problem?
No not really. There has been some discussion about this on the mailing list. The result is:
Build your own Set/Map abstraction on top of Set/Map, which would convert the objects to a primitive value according to your hashing function.
Wait for value objects coming in ES7.

This will do it for constructors like what you're working with.
var sameInstance = function(obj1, obj2){
return obj2 instanceof ob1.constructor;
};
Some other types of objects you might need something different, but this should be ok for what you need.
The above is what you need in function form. To get it to work with Set you will have to inherit the Set object, and override the has method with your own has.

Related

Javascript this.function vs return function?

Im a bit confused learning all the new ES6 vs ES5 syntax with javascript and when it comes to functions/classes with methods and calling upon those methods I can't really tell what is the "right way".
For example take this code:
function account(amount) {
let cash = amount;
this.setCash = function(amt)
{
cash = amt;
}
this.getCash = function()
{
return cash;
}
}
var person = new account(100);
console.log(person.getCash()); //returns 100
person.setCash(150);
console.log(person.getCash()); //returns 150
Works like normal as expected (This is how I originally saw the methods being used when going through tutorials).
However i've seen this occasionally:
function account2(amount) {
let cash = amount;
function setCash(amt)
{
cash = amt;
}
function getCash()
{
return cash;
}
return{
getCash,
setCash
}
}
var person2 = new account2(50);
console.log(person2.getCash()); //returns 50
person2.setCash(175);
console.log(person2.getCash()); //returns 175
Both of these work perfectly fine, and do as I think they should. However is one just an older way of doing it? or less correct maybe? This is my biggest barrier in learning JS right now, since ES6 is here there are so many different ways to do something as simple as making a "class" in JS with methods.
Personally the first way seems easier since you don't have to return the functions....but for example at work I see the 2nd way used mostly?
If you removed new from the second version (more on that in a moment) then both of your examples are perfectly fine ways of creating an object with private data accessed via public get/set methods. Which one you would choose is personal preference, or dependent on whether you want to extend the functionality further with prototype methods, etc.
The reason I suggest removing new from the second version is that although both examples work, the first is a "normal" use of the new operator, but the second is an "incorrect" use of the new operator - not a syntax error, just semantically incorrect and potentially misleading for other people reading your code.
So why is that second use of new semantically incorrect? When you say var person = new account(100):
A new object is created with a prototype that is account.prototype. That means the new object inherits any methods and properties of the account.prototype, and from its prototype, etc. (In your case you haven't defined any methods on the prototype, but you could.)
Within the account() function, this will refer to the object created in step 1. Which is why you can attach methods to it with this.setCash = ...
The new object is returned from account() by default. This step will not occur if there is an explicit return statement returning something else as in your account2() function.
So in the first example, person instanceof account is true, because the default object was returned. In the second example, person2 instanceof account2 is false, because a different object was returned - this is misleading for people reading your code, because when they see person2 = new account2() they might reasonably expect that person2 will be an instance of account2 but it's not. It's also a bit inefficient, because the JS engine goes to the trouble of creating an object for you and then you don't use it.
So the second example would be more "correct" without new, providing the same behaviour minus the unnecessary auto object creation:
var person2 = account2(50);
(Almost) any JS function that you write can be called with new, but there is a convention that we describe functions intended to be called with new as "constructors", and usually the function name is capitalised, so Account(), not account(). (The spelling doesn't change the behaviour, it's just an extra hint for people reading the code.)
Note that use of new isn't the only way to link objects to particular prototypes: it is the old-school-but-still-perfectly-valid way, but you can use Object.create() instead.
By the way, the behaviour in question is not new in ES5 or 6, though let and the shortcut object literal syntax in account2 are new.
When you create an instance with new keyword this is returned implicitly where as when assign a function to a variable without new, you need to use return explicitly.

Best practice when adding custom method to Array (Built-in object)

I have created a node module with a couple of custom methods for arrays and strings.
First I just used it like a regular module and got the functions from a require like this:
Alt 1.
const invSlice = require('inverted-slice');
let arr1 = [1,2,3,4];
invSlice.iSlice(arr, start, stop);
This works but it would be nicer to call iSlice as a method on the Array object. I solved this by adding the following code in my library:
Array.prototype.iSlice = iSliceBuiltin; // iSliceBuiltin is my function
And the method can now be used like:
Alt 2.
require('inverted-slice');
let arr1 = [1,2,3,4];
arr1.iSlice(start, stop);
Which I think is nicer then Alt 1.
Question
My question is if there is any best practice or guidelines to follow when adding custom methods like in Alt 2 to built-in Objects like Array or String ?
Extending built-in prototypes has always triggered debates, and I think we can conclude it is not considered best practice.
On the other hand it is indeed nice if you can call these custom methods as object methods instead of plain functions.
You might consider a wrapper function that will return an Array instance that has the extra methods defined for it: i.e., not on the prototype, but on the Array instance itself.
Your module could look like this:
function iArray(arr) {
return Object.assign([], arr || [], {
iSlice: iSliceBuiltin,
iSplice: iSpliceBuiltin
});
}
// ... your module functions come here, but excluding the changes to the Array prototype
module.exports = {
iArray
}
Then you would use it like this:
const iArray = require('inverted-slice');
let arr1 = iArray([1,2,3,4]); // enrich array with extra methods
let result = arr1.iSlice(0, 1);
To allow chaining, you could change the return statement in iSliceSpliceHelper to:
return iArray(newArr);
So, now you can write:
let arr1 = iArray([1,2,3,4]); // enrich array with extra methods
let result = arr1.iSlice(0, 1).iSlice(1, 2);
Existing libraries might implement your alternative 1 (e.g. underscore), but many also go for something like I propose here. See for instance Sugar (new Sugar.Array([1,2,3])), or Lazy (Lazy([1,2,3])).
In small doses I think it's not that big of a deal to use Alt 2, but I believe that over usage can create problems. If I remember correctly, they had to completely redo Cut The Rope due to performance problems that I believe stemmed largely in part from prototype extensions. You may also want to consider posting this on https://codereview.stackexchange.com/
A couple references:
http://perfectionkills.com/whats-wrong-with-extending-the-dom/
https://softwareengineering.stackexchange.com/questions/104320/why-is-extending-the-dom-built-in-object-prototypes-a-bad-idea

inherit from Javascript Array or typed array

I want to make a JavaScript class vector which is a zero-initialized array. I'll probably want to add math functionality later, but I don't want to sacrifice the memory or performance qualities of the native types because the program operates on a lot of data. (It's basically scientific visualization.)
To insert Array.prototype in the prototype chain, I tried to use
vector.prototype = Object.create( Array.prototype );
Firefox gives me an error
TypeError: Array.prototype.toSource called on incompatible Object
Is this a bug in Firefox? It seems to work in Webkit.
So, I tried to use Float32Array which is closer to what I want, and zero-initialized by default anyway.
var vector = function( size ) { Float32Array.call( this, size ); }
vector.prototype = Object.create( Float32Array.prototype );
In Firefox, this runs but new doesn't initialize objects properly. In Webkit new vector throws an exception. Makes no difference if I use vector.prototype = Float32Array.prototype instead.
Am I fundamentally asking for too much?
It is nearly impossible to truly inherit from the JavaScript Array because of very special functionality that is built in to the language and how it handles arrays. You will see a ton of answers but most of them will fail under special circumstances.
1) if you inherit it, it will lose the special length properties.
2) if you use the __ ProtoType __ variable, it is non standard and will not work on all browsers.
3) There are many ways of detecting arrays in 3rd party libraries that might fail when checked. instanceof Array.isArray and ==['Object Array'] could all be used.
For more detailed reading, I would suggest this article How ECMA still does not allow to subclass array
I have a solution of returning a real array and avoiding all of these issues.
Here is the fiddle Inhetit from JavaScript Array
There are detection, length and functionality tests on the fiddle.
var ArrayGenerator = (function () {
var args = [].slice.call(arguments);
var Arr = new Array();
var ovevar ArrayGenerator = (function () {
var args = [].slice.call(arguments);
var Arr = new Array();
var overiddenPush = Arr.push;
args.forEach(function (arg) {
Arr.push(arg);
});
args = null;
Arr.push = function (args) {
overiddenPush.apply(this, arguments);
}
Arr.TableName = "YourTable";
return Arr;
});
//You initialize your array like this
var MyArray=new ArrayGenerator(1,2,3);
Returns Real Array
No Detection Issues
Allows you to override everything you want and add custom properties and methods just like an inherited array.
You simply create an array, store the original methods in internal variables and replace the public methods with you own. To call the base, you simple call your referenced methods instead of prototype like a regular inherited object.

Making an associative array in Javascript that has custom get/put operations

I need to make a Javascript object that would behave as an associative array, but with some functions that are called before getting and setting properties.
For example, the task may be like this: we should make an object, that would contain a squared value of a key, like this:
obj.two should be equal to 4,
obj.four should be equal to 16,
obj['twenty one'] should be equal to 441.
This is an example. Actually I need to make setting operation overridden too. The getting and setting operations would go to the database, and they not necceserily would take strings as keys, but any types of objects, from which it would create a DB query.
How would I do that a) with as less thirdparty libraries as possible and b) to make it work on as much platforms as possible?
I am new to JS, I've found that JS has no associative arrays, relying on the ability to define objects on the fly with arbitrary properties. I googled and had an idea to use or even override lookupgetter (and setter), where define a new getter/setter on the fly, but I coundn't find if the interpreter would use this method every time it encounters new key. Anyway, it looks like I wouldn't be able to use anything except strings or maybe numbers as keys.
In Java, I would just implement java.util.Map.
Please help me, how would I do the same in Javascript?
edit
I think I will get what I want if I manage to override [[Get]] and [[Put]] methods mentioned here http://interglacial.com/javascript_spec/a-8.html#a-8.6.2.1
For your example, doesn't this do what you want:
var myObj = {};
myObj["two"] = 4;
myObj["four"] = 16;
myObj["twenty one"] = 441;
alert(myObj["four"]); // says 16
Or are you trying to say that the object should magically calculate the squares for you?
JavaScript object keys are strings. If you try to use a number as a key JavaScript basically converts it to a string first.
Having said that, you can use objects as keys if you define a meaningful toString method on them. But of course meaningful is something that happens on a case by case basis and only you will know what needs to be done for your case.
You can also define objects that maintain their own internal data structures which you access via object methods. I think explaining that is beyond the scope of this post. Google "javascript module pattern" for some pointers to get you started.
See http://ejohn.org/blog/javascript-getters-and-setters/
Also this particular answer: Javascript getters and setters for dummies?
edit
According to Does JavaScript have the equivalent of Python's __getattribute__? and Is there an equivalent of the __noSuchMethod__ feature for properties, or a way to implement it in JS? there is no nice way of accomplishing exactly what the OP wants. Getters and setters are not useful because you must know the name of what you're looking for in advance.
My recommendation would thus be to do something like:
var database = {}
database.cache = {}
database.get = function(key) {
// INSERT CUSTOM LOGIC to recognize "forty-two"
if (!(key in database.data))
database.cache[key] = fetch_data_from_database();
return database.cache[key];
}
database.put = function(key, value) {
database.cache[key] = value;
send_data_to_database(key, value);
}
I decided that the most correct way to implement this is to use Harmony:Proxies. It isn't working on all platforms, but it lets implement this in the most seamless way; and it may be supported in more platforms in the future.
This page contains an example that I used as a template to do what I want:
http://wiki.ecmascript.org/doku.php?id=harmony:proxies

workaround: javascript dictionary which takes objects as keys

I read a few questions and answers about javascript dictionary implementations, but they don't meet my requirements:
the dictionary must be able to take objects as keys
the values must be accessible by the []-operator
So I came up with the idea to overwrite the valueOf-method in Object.prototype, as follows:
Object.__id__ = 0;
Object.prototype.valueOf = function() {
if(!this.__id__)
this.__id__ = ++Object.__id__;
return "__id__" + this.__id__;
}
Object.prototype.toString = Object.prototype.valueOf;
//test
var x = {p1: "5"};
var y = [6];
var z = {};
z[x] = "7";
z[y] = "8";
console.log(z[x], z[y]);
I tested this with google-chrome and it seems to work well, but I'm a bit sceptical, whether this will cause some drawbacks, since it was so easy to implement.
Considering that the valueOf method is not used for other purposes in the whole code, do you think there are any disadvantages?
It's an interesting idea. I suggest my jshashtable. It meets your first requirement but not the second. I don't really see the advantage of insisting on using the square bracket property access notation: do you have a particular requirement for it?
With jshashtable, you can provide a hashing function to the Hashtable constructor. This function is passed an object to be used as a key and must return a string; you could use a function not dissimilar to what you have there, without having to touch Object.prototype.
There are some disadvantages to your idea:
Your valueOf method will show up in a for...in loop over any native object;
You have no way determining which keys should be considered equal, which is something you may want to do. Instead, all keys will be considered unique.
This won't work with host objects (i.e. objects provided by the environment, such as DOM elements)
It is an interesting question, because I had so far assumed that any object can be used as an index (but never tried with associative arrays). I don't know enough about the inner workings of JavaScript to be sure, but I'd bet that valueOf is used somewhere else by JavaScript, even if not in your code. You might run into seemingly inexplicable problems later. At least, I'd restrict myself to a new class and leave Object alone ;) Or, you explicitly call your hashing function, calling it myHash() or whatever and calling z[x.myHash()] which adds clutter but would let me, personally, sleep better ;) I can't resist thinking there's a more JavaScript-aware solution to this, so consider all of these ugly workarounds ;)
If you came upon this question looking for a JS dictionary where objects are keys look at Map Map vs Object in JavaScript

Categories

Resources