Reputation: 79358
After looking at a simple hash table implementation in JavaScript, the key index is computed as:
function index(str, max) {
var hash = 0;
for (var i = 0; i < str.length; i++) {
var letter = str[i];
hash = (hash << 5) + letter.charCodeAt(0);
hash = (hash & hash) % max;
}
return hash;
}
So I'm wondering in the case of v8, how it uses a function similar to that but makes sure the index is unique on the object. So if you do this:
{ a: 'foo', b: 'bar' }
Then it becomes something like:
var i = index('a', 100000)
// 97
var j = index('b', 100000)
// 98
But if you have 100's or 1000's or more keys on an object, it seems like there might be collisions.
Wondering how a hashtable guarantees they are unique, using v8 as a practical example.
Upvotes: 1
Views: 306
Reputation: 40561
V8 developer here. Hashes of strings are not unique (that's kind of the point of using a hash function); V8 uses quadratic probing to deal with collisions (see source). You can read more about various strategies at https://en.wikipedia.org/wiki/Hash_table#Collision_resolution.
Also, hash = (hash & hash) % max;
is pretty silly ;-)
Upvotes: 3