I always assumed caching the length of an array in JavaScript is a good idea (especially in the condition of a for
loop) because of the expensiveness of calculating the length of an array.
Example
for (var i = 0; i < arr.length; i++) { }
// vs
for (var i = 0, arrLength = arr.length; i < arrLength; i++) { }
However, I thought perhaps the length
property is only updated on creation and alteration of the array. Therefore, reading it shouldn't be too expensive an operation as opposed to reading it stored in a variable (as opposed to other methods in other languages that may need to seek in memory to find the end of something, e.g. strlen()
in C).
I have two questions. I am also interested in how this works, so please don't hit me with the premature optimisation stick.
Assume the JavaScript engi开发者_如何学Gones in browsers.
- Is there any advantage to caching the
length
property of an array in JavaScript? Is there much more involved in reading a local variable over an object's property? - Is the
length
property simply altered on creation and onshift()
andpop()
type methods that don't return a new array and otherwise simply stored as an integer?
Well, I would have said it was expensive, but then I wrote a little test @ jsperf.com and to my surprise using i<array.length
actually was faster in Chrome, and in FF(4) it didn't matter.
My suspicion is that length is stored as an integer (Uint32). From the ECMA-specs (262 ed. 5, page 121):
Every Array object has a length property whose value is always a nonnegative integer less than 232. The value of the length property is numerically greater than the name of every property whose name is an array index; whenever a property of an Array object is created or changed, other properties are adjusted as necessary to maintain this invariant. Specifically, whenever a property is added whose name is an array index, the length property is changed, if necessary, to be one more than the numeric value of that array index; and whenever the length property is changed, every property whose name is an array index whose value is not smaller than the new length is automatically deleted. This constraint applies only to own properties of an Array object and is unaffected by length or array index properties that may be inherited from its prototypes
Phew! I don't know if I ever get used to such language ...
Finally, we always have our good old lagging behind browser. In IE (9, 8, 7) caching the length is really faster. One of many more reasons to not use IE, I say.
TL;DR:
From what I can gather, it seems like the length of the array is cached internally (at least in V8)..
(Details? Read on :))
So, this question has dinged around in my head a few times and I've decided to get to the root of the problem (at least, in one implementation).
Digging around V8 source yielded the JSArray class.
// The JSArray describes JavaScript Arrays
// Such an array can be in one of two modes:
// - fast, backing storage is a FixedArray and length <= elements.length();
// Please note: push and pop can be used to grow and shrink the array.
// - slow, backing storage is a HashTable with numbers as keys.
I'm making an assumption that the type of array elements dictates whether it's fast or slow. I got down to a bit flag being set in set_has_fast_elements
(set_bit_field2(bit_field2() | (1 << kHasFastElements))
), which is where I figured I'd draw the digging line as I was looking in google code and don't have the source locally.
Now, it seems that any time any operation is done on the array (which is a child class of JSObject
, a call is made to NormalizeElements()
, which executes the following:
// Compute the effective length.
int length = IsJSArray() ?
Smi::cast(JSArray::cast(this)->length())->value() :
array->length();
So, in answering your questions:
- There doesn't seem to be any advantage in Chrome (or other browsers that use V8) to caching the
length
property of an array (unless you're doing something odd that would force it to beslow
(I'm not sure what those conditions are) - having said that, I'll most likely continue to cachelength
until I get a chance to go through all OS browser implementations ;) - The
length
property seems to be altered after any operation on the object.
Edit:
On a side note, it seems that an "empty" array is actually allocated to have 4 elements:
// Number of element slots to pre-allocate for an empty array.
static const int kPreallocatedArrayElements = 4;
I'm not sure how many elements the array grows by once the bounds have been exceeded - I didn't dig that deep :)
Another set of performance tests. The loop is done over an array of million random numbers with an empty loop.
In Chrome, the loops with cached and non-cached lengths clock pretty much the same times, so I am guessing it's a V8 optimization to cache the length.
In Safari and Firefox, the cached length was consistently about 2x faster than the non-cached version.
This article investigates automatic caching in V8 and Chrome, by asking IRHydra for the generated code:
How the Grinch stole array.length access by Vyacheslav Egorov
He found that under certain conditions manually caching the .length
actually added overhead rather than improving performance!
But anyway, this kind of micro-optimization is unlikely to achieve any noticeable gain for your users. For their benefit, and for yours, focus instead on code that is clear to read, and using good data structures and algorithms in your code!
Avoid premature optimisation: Focus on elegant code until a performance issue arises. Only then, seek out the bottleneck through profiling, and then optimise just that part of the code.
Just a note:
On some browsers (I've noticed it in Safari, IE, and Opera), you can get a speed boost by caching the length inside the for loop declaration:
var j;
for (var i = 0, len = arr.length; i < len; i++) {
j = arr[i];
}
I edited @KooiInc's jsperf test above to add this case.
Take care not to assume that this is true for all iterable collections. For example caching the length of an HTMLCollection is 65% faster in Chrome (Version 41) and 35% faster in Firefox (Version 36).
http://jsperf.com/array-length-in-loop-dom
精彩评论