Reputation: 12705
Most of the tutorials that I've read on arrays in JavaScript (including w3schools and devguru) suggest that you can initialize an array with a certain length by passing an integer to the Array constructor using the var test = new Array(4);
syntax.
After using this syntax liberally in my js files, I ran one of the files through jsLint, and it freaked out:
Error: Problem at line 1 character 22: Expected ')' and instead saw '4'.
var test = new Array(4);
Problem at line 1 character 23: Expected ';' and instead saw ')'.
var test = new Array(4);
Problem at line 1 character 23: Expected an identifier and instead saw ')'.
After reading through jsLint's explanation of its behavior, it looks like jsLint doesn't really like the new Array()
syntax, and instead prefers []
when declaring arrays.
So I have a couple questions:
First, why? Am I running any risk by using the new Array()
syntax instead? Are there browser incompatibilities that I should be aware of?
And second, if I switch to the square bracket syntax, is there any way to declare an array and set its length all on one line, or do I have to do something like this:
var test = [];
test.length = 4;
Upvotes: 962
Views: 1493800
Reputation: 160
JavaScript pretty much has 2 types of array entries, namely mappable
and unmappable
.
Both entry types are iterable
, meaning they both work with the for-of
loop.
A JavaScript array can contain both mappable
and unmappable
array entries.
To prove this I present the following code sample:
const a = (Array(1)).concat([...Array(1)]).concat(['literal']).concat(Array(1));
console.log('a:', a.length, a);
a: 4 [ <1 empty item>, undefined, 'literal', <1 empty item> ]
The same mostly works with the .push()
method:
const b = (Array(1));
b.push(...[...Array(1)]);
b.push(['literal']);
b.push(...Array(1));
console.log('b:', b.length, b);
b: 4 [ <1 empty item>, undefined, 'literal', undefined ]
It also mostly works with the .unshift()
method:
const c = (Array(1));
c.unshift(...[...Array(1)]);
c.unshift(['literal']);
c.unshift(...Array(1));
console.log('c:', c.length, c);
c: 4 [ undefined, 'literal', undefined, <1 empty item> ]
NOTE that spreading (...
) an unmappable
array produces a tuple of undefined
entries, which is why I've said that the above mostly works the same as with the .concat()
method.
unmappable
array entries:Empty-Item
-Filled Arrays ( Length = Given ):const myArray = Array(LENGTH);
//------------OR------------//
const myArray = new Array(LENGTH);
console.log('myArray:', myArray.length, myArray);
myArray: LENGTH [ <LENGTH empty items> ]
Empty-Item
here refers to the inherently unmappable
type of entry that the Array()
builtin fills arrays with.
NOTE that Array()
and new Array()
do the exact same thing in JavaScript. So, there really isn't a good reason to use the new
keyword with the Array
builtin.
mappable
array entries:Undefined
-Filled Arrays ( Length = Given ):const myArray = Array(LENGTH).fill();
//------------OR------------//
const myArray = Array.from({ length: LENGTH });
//------------OR------------//
const myArray = [...Array(LENGTH)];
//------------OR------------//
const myArray = [...(new Array(LENGTH))];
console.log('myArray: ', myArray.length, myArray);
myArray: LENGTH [ undefined, undefined, ___ ]
The new Array()
version is just mentioned here for completeness sake.
Surrogate
-Filled Arrays ( Length = Given ):const myArray = Array(LENGTH).fill(SURROGATE_VALUE);
console.log('myArray:', myArray.length, myArray);
myArray: LENGTH [ SURROGATE_VALUE , SURROGATE_VALUE, ___ ]
The SURROGATE_VALUE
is often chosen as 0
for mathematical use, but it can really be anything.
WARNING: As noted in earlier asnwers, the .fill()
method doesn't copy the result passed to it. So, reference data types, like Object
and Array
, should NOT be used as SURROGATE_VALUE
s.
If you need a deeply nested array, initialized with reference data types, like Object
and Array
, as entries, then it's best to start with a mappable
array initialized with undefined
entries and to .map()
each undefined
entry to a new instance of the needed reference data type.
const myArray = [SOME_VALUE_A, SOME_VALUE_B, ___];
console.log('myArray:', myArray.length, myArray);
myArray: IMPLICIT_LENGTH [ SOME_VALUE_A, SOME_VALUE_B, ___ ]
Here the length
of the array is implicit since it is equal to the number of meaningful entries you've initialized it with.
0
):const myArray = [];
//------------OR------------//
const myArray = Array();
//------------OR------------//
const myArray = new Array();
console.log('myArray: ', myArray.length, myArray);
myArray: 0 []
These arrays contain neither mappable
nor unmappable
entries since they are empty. So, they are classified separately to make it clear that it doesn't matter how you initialize an empty array.
Unmappable
and Mappable
Entries:Unmappable
entries stay the same, whilst mappable
entries are mapped.
const a = (Array(1)).concat([...Array(1)]).concat(['literal']).concat(Array(1));
const m = a.map(() => true);
console.log('a:', a.length, a);
console.log('m:', m.length, m);
a: 4 [ <1 empty item>, undefined, 'literal', <1 empty item> ]
m: 4 [ <1 empty item>, true, true, <1 empty item> ]
As you can see, the empty-item
or unmappable
entries stay the same, whilst the mappable
entries, which in the example include an entry of undefined
and a string literal, are both mapped to the desired value.
Unmappable
and Mappable
Entries:const a = (Array(1)).concat([...Array(1)]).concat(['literal']).concat(Array(1));
for (let i = 0; i < a.length; i++) {
console.log(i, a[i]);
}
0 undefined
1 undefined
2 literal
3 undefined
Similar to the spread operator (...
) and with the .push()
and .unshift()
methods, empty-item
entries take on the value of undefined
when accessed via the array bracket accessor myArray[INDEX]
notation.
for-of
loops:const a = (Array(1)).concat([...Array(1)]).concat(['literal']).concat(Array(1));
for (const entry of a) {
console.log(entry);
}
undefined
undefined
literal
undefined
Upvotes: 3
Reputation: 2918
The simplest form is to use
Array.from({ length: 3 });
// gives you
[undefined, undefined, undefined]
Unlike Array(3)
which will give you an array you can't iterate over. Array.from({ length })
gives you an array you can iterate easily.
Array.from({ length: 3 }).map((e, idx) => `hi ${idx}`);
// ['hi 1', 'hi 2', 'hi 3']
Upvotes: 31
Reputation: 1348
In addition to the answers of others, another clever way is to use Float32Array
to create an array and iterate on it.
For this purpose, create an instance from Float32Array
with your desired length like this:
new Float32Array(5)
This code returns an array-like that you can convert it to an array with Array.from()
:
Array.from(new Float32Array(5)) // [0, 0, 0, 0, 0]
You can also use fill()
to change the value of items:
Array.from(new Float32Array(5).fill(2)) // [2, 2, 2, 2, 2]
And of course you can iterate on it:
Array.from(new Float32Array(5)).map(item => /* ... */ )
Upvotes: 3
Reputation: 24508
In modern JS engines, sparse arrays are fully supported. You can use []
or new Array(len)
in any way you like, even with random access. Dictionary mode seems to be a thing of the past.
In current Chrome (and I guess any V8 environment), Arrays can have a length of up to 2^32-1 and allocation is sparse (meaning empty chunks don't use up any memory):
On the one hand, for
loops work as intended, however, Array
's builtin higher order functions (such as map
, filter
, find
, some
etc.) ignore unassigned elements. They require fill
(or some other method of population) first:
const a = new Array(10);
const b = new Array(10).fill(0);
a.forEach(x => console.log(x)); // does nothing
b.forEach(x => console.log(x)); // works as intended
(I removed most of the old version.) The gist was that creating a large array using new Array(largeNumber)
or random accessing an array in places that have not yet been allocated would tumble it into "dictionary mode". Meaning you are using an array with indexes, but under the hood it would use a dictionary to store the values, thus messing with performance, and also with iteration behavior. Luckily that is a thing of the past.
Upvotes: 64
Reputation: 4176
[...Array(6)].map(x => 0);
// [0, 0, 0, 0, 0, 0]
OR
Array(6).fill(0);
// [0, 0, 0, 0, 0, 0]
Note: you can't loop empty slots i.e. Array(4).forEach(() => …)
OR
( typescript safe )
Array(6).fill(null).map((_, i) => i);
// [0, 1, 2, 3, 4, 5]
OR
Classic method using a function ( works in any browser )
function NewArray(size) {
var x = [];
for (var i = 0; i < size; ++i) {
x[i] = i;
}
return x;
}
var a = NewArray(10);
// [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
When creating a 2D array with the fill
intuitively should create new instances. But what actually going to happen is the same array will be stored as a reference.
var a = Array(3).fill([6]);
// [ [6], [6], [6] ]
a[0].push(9);
// [ [6, 9], [6, 9], [6, 9] ]
Solution
var a = [...Array(3)].map(x => []);
a[0].push(4, 2);
// [ [4, 2], [], [] ]
So a 3x2 Array will look something like this:
[...Array(3)].map(x => Array(2).fill(0));
// [ [0, 0], [0, 0], [0, 0] ]
function NArray(...dimensions) {
var index = 0;
function NArrayRec(dims) {
var first = dims[0], next = dims.slice().splice(1);
if(dims.length > 1)
return Array(dims[0]).fill(null).map((x, i) => NArrayRec(next ));
return Array(dims[0]).fill(null).map((x, i) => (index++));
}
return NArrayRec(dimensions);
}
var arr = NArray(3, 2, 4);
// [ [ [ 0, 1, 2, 3 ] , [ 4, 5, 6, 7] ],
// [ [ 8, 9, 10, 11] , [ 12, 13, 14, 15] ],
// [ [ 16, 17, 18, 19] , [ 20, 21, 22, 23] ] ]
var Chessboard = [...Array(8)].map((x, j) => {
return Array(8).fill(null).map((y, i) => {
return `${String.fromCharCode(65 + i)}${8 - j}`;
});
});
// [ [A8, B8, C8, D8, E8, F8, G8, H8],
// [A7, B7, C7, D7, E7, F7, G7, H7],
// [A6, B6, C6, D6, E6, F6, G6, H6],
// [A5, B5, C5, D5, E5, F5, G5, H5],
// [A4, B4, C4, D4, E4, F4, G4, H4],
// [A3, B3, C3, D3, E3, F3, G3, H3],
// [A2, B2, C2, D2, E2, F2, G2, H2],
// [A1, B1, C1, D1, E1, F1, G1, H1] ]
handy little method overload when working with math
function NewArray( size , method, linear )
{
method = method || ( i => i );
linear = linear || false;
var x = [];
for( var i = 0; i < size; ++i )
x[ i ] = method( linear ? i / (size-1) : i );
return x;
}
NewArray( 4 );
// [ 0, 1, 2, 3 ]
NewArray( 4, Math.sin );
// [ 0, 0.841, 0.909, 0.141 ]
NewArray( 4, Math.sin, true );
// [ 0, 0.327, 0.618, 0.841 ]
var pow2 = ( x ) => x * x;
NewArray( 4, pow2 );
// [ 0, 1, 4, 9 ]
NewArray( 4, pow2, true );
// [ 0, 0.111, 0.444, 1 ]
Upvotes: 394
Reputation: 1597
The shortest:
let arr = [...Array(10)];
console.log(arr);
Upvotes: 100
Reputation: 44316
In most answers it is recommended to fill
the array because otherwise "you can't iterate over it", but this is not true. You can iterate an empty array, just not with forEach
. While loops, for of loops and for i loops work fine.
const count = Array(5);
Does not work.
console.log('---for each loop:---');
count.forEach((empty, index) => {
console.log(`counting ${index}`);
});
These work:
console.log('---for of loop:---');
for (let [index, empty] of count.entries()) {
console.log(`counting for of loop ${index}`);
}
console.log('---for i loop:---');
for (let i = 0, il = count.length; i < il; ++i) {
console.log(`counting for i loop ${i}`);
}
console.log('---while loop:---');
let index = 0;
while (index < count.length) {
console.log(`counting while loop ${index}`);
index++;
}
Check this fiddle with the above examples.
Also angulars *ngFor
works fine with an empty array:
<li *ngFor="let empty of count; let i = index" [ngClass]="
<span>Counting with *ngFor {{i}}</span>
</li>
Upvotes: 1
Reputation: 1448
Assuming that Array's length is constant. In Javascript, This is what we do:
const intialArray = new Array(specify the value);
Upvotes: 4
Reputation: 791
Please people don't give up your old habits just yet. There is a large difference in speed between allocating memory once then working with the entries in that array (as of old), and allocating it many times as an array grows (which is inevitably what the system does under the hood with other suggested methods).
None of this matters of course, until you want to do something cool with larger arrays. Then it does.
Seeing as there still seems to be no option in JS at the moment to set the initial capacity of an array, I use the following...
var newArrayWithSize = function(size) {
this.standard = this.standard||[];
for (var add = size-this.standard.length; add>0; add--) {
this.standard.push(undefined);// or whatever
}
return this.standard.slice(0,size);
}
There are tradeoffs involved:
standard
array does permanently reserve as much space as the largest array you have asked for.But if it fits with what you're doing there can be a payoff. Informal timing puts
for (var n=10000;n>0;n--) {var b = newArrayWithSize(10000);b[0]=0;}
at pretty speedy (about 50ms for the 10000 given that with n=1000000 it took about 5 seconds), and
for (var n=10000;n>0;n--) {
var b = [];for (var add=10000;add>0;add--) {
b.push(undefined);
}
}
at well over a minute (about 90 sec for the 10000 on the same chrome console, or about 2000 times slower). That won't just be the allocation, but also the 10000 pushes, for loop, etc..
Upvotes: 9
Reputation: 8911
With ES2015 .fill()
you can now simply do:
// `n` is the size you want to initialize your array
// `0` is what the array will be filled with (can be any other value)
Array(n).fill(0)
Which is a lot more concise than Array.apply(0, new Array(n)).map(i => value)
It is possible to drop the 0
in .fill()
and run without arguments, which will fill the array with undefined
. (However, this will fail in Typescript)
Upvotes: 610
Reputation: 2800
ES6 introduces Array.from
which lets you create an Array
from any "array-like" or iterables objects:
Array.from({length: 10}, (x, i) => i);
// [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
In this case {length: 10}
represents the minimal definition of an "array-like" object: an empty object with just a length
property defined.
Array.from
allows for a second argument to map over the resulting array.
Upvotes: 68
Reputation: 13294
Array(5)
gives you an array with length 5 but no values, hence you can't iterate over it.
Array.apply(null, Array(5)).map(function () {})
gives you an array with length 5 and undefined as values, now it can be iterated over.
Array.apply(null, Array(5)).map(function (x, i) { return i; })
gives you an array with length 5 and values 0,1,2,3,4.
Array(5).forEach(alert)
does nothing, Array.apply(null, Array(5)).forEach(alert)
gives you 5 alerts
ES6
gives us Array.from
so now you can also use Array.from(Array(5)).forEach(alert)
If you want to initialize with a certain value, these are good to knows...
Array.from('abcde')
, Array.from('x'.repeat(5))
or Array.from({length: 5}, (v, i) => i) // gives [0, 1, 2, 3, 4]
Upvotes: 1183
Reputation: 17
You can set the array length by using array.length = youValue
So it would be
var myArray = [];
myArray.length = yourValue;
Upvotes: -2
Reputation: 48346
Here is another solution
var arr = Array.apply( null, { length: 4 } );
arr; // [undefined, undefined, undefined, undefined] (in Chrome)
arr.length; // 4
The first argument of apply()
is a this object binding, which we don't care about here, so we set it to null
.
Array.apply(..)
is calling the Array(..)
function and spreading out the { length: 3 }
object value as its arguments.
Upvotes: 11
Reputation: 816232
Why do you want to initialize the length? Theoretically there is no need for this. It can even result in confusing behavior, because all tests that use the length
to find out whether an array is empty or not will report that the array is not empty.
Some tests show that setting the initial length of large arrays can be more efficient if the array is filled afterwards, but the performance gain (if any) seem to differ from browser to browser.
jsLint does not like new Array()
because the constructer is ambiguous.
new Array(4);
creates an empty array of length 4. But
new Array('4');
creates an array containing the value '4'
.
Regarding your comment: In JS you don't need to initialize the length of the array. It grows dynamically. You can just store the length in some variable, e.g.
var data = [];
var length = 5; // user defined length
for(var i = 0; i < length; i++) {
data.push(createSomeObject());
}
Upvotes: 500
Reputation: 533
I'm surprised there hasn't been a functional solution suggested that allows you to set the length in one line. The following is based on UnderscoreJS:
var test = _.map(_.range(4), function () { return undefined; });
console.log(test.length);
For reasons mentioned above, I'd avoid doing this unless I wanted to initialize the array to a specific value. It's interesting to note there are other libraries that implement range including Lo-dash and Lazy, which may have different performance characteristics.
Upvotes: 15
Reputation: 5303
(this was probably better as a comment, but got too long)
So, after reading this I was curious if pre-allocating was actually faster, because in theory it should be. However, this blog gave some tips advising against it http://www.html5rocks.com/en/tutorials/speed/v8/.
So still being unsure, I put it to the test. And as it turns out it seems to in fact be slower.
var time = Date.now();
var temp = [];
for(var i=0;i<100000;i++){
temp[i]=i;
}
console.log(Date.now()-time);
var time = Date.now();
var temp2 = new Array(100000);
for(var i=0;i<100000;i++){
temp2[i] = i;
}
console.log(Date.now()-time);
This code yields the following after a few casual runs:
$ node main.js
9
16
$ node main.js
8
14
$ node main.js
7
20
$ node main.js
9
14
$ node main.js
9
19
Upvotes: 8
Reputation: 46735
The array constructor has an ambiguous syntax, and JSLint just hurts your feelings after all.
Also, your example code is broken, the second var
statement will raise a SyntaxError
. You're setting the property length
of the array test
, so there's no need for another var
.
As far as your options go, array.length
is the only "clean" one. Question is, why do you need to set the size in the first place? Try to refactor your code to get rid of that dependency.
Upvotes: 3
Reputation: 185873
This will initialize the length property to 4:
var x = [,,,,];
Upvotes: 30