I think one weakness of maps and sets is that an object can have a hash or compareTo to live in the container but also other members not part of the key. These objects should be searchable, updatable, even key updatable such that they are relocated in the container.
Sure, hash is cool, the main cost being the hash function itself. The main variance is whether the bucket is a list to search or if you jump to another bucket on miss with quadratic, linear, double hash. The main control is the initial bucket count. Some hash maps/sets can expand, doubling in bucket count, but cleanup is a challenge, so going big initially is wise. Buckets are often pointers, so 8 bytes each is pretty cheap!
Do you have an opinion about bucket count, like is a prime better than a mod 2 number for randomizing buckets?
The trade off in choosing hash is lookup speed versus lack of ordering. In comparison, a tree or trie allows range search, prefix search, sorted output but a bit slower lookup, tree imbalance vulnerability, a bit more memory cost. Linked list, tree, trie allows no danger of too high a load factor, infinite expansion. All but array have extra overhead to size. Array, vector are fast but expensive to expand, insert, delete, sort. Skip list combines tree and linked list! So many choices!
For further actions, you may consider blocking this person and/or reporting abuse
We're a place where coders share, stay up-to-date and grow their careers.
I think one weakness of maps and sets is that an object can have a hash or compareTo to live in the container but also other members not part of the key. These objects should be searchable, updatable, even key updatable such that they are relocated in the container.
What you have pointed out is somewhat very advanced issue
But my general focus here was how it can help in tech interviewes
Sure, hash is cool, the main cost being the hash function itself. The main variance is whether the bucket is a list to search or if you jump to another bucket on miss with quadratic, linear, double hash. The main control is the initial bucket count. Some hash maps/sets can expand, doubling in bucket count, but cleanup is a challenge, so going big initially is wise. Buckets are often pointers, so 8 bytes each is pretty cheap!
Do you have an opinion about bucket count, like is a prime better than a mod 2 number for randomizing buckets?
The trade off in choosing hash is lookup speed versus lack of ordering. In comparison, a tree or trie allows range search, prefix search, sorted output but a bit slower lookup, tree imbalance vulnerability, a bit more memory cost. Linked list, tree, trie allows no danger of too high a load factor, infinite expansion. All but array have extra overhead to size. Array, vector are fast but expensive to expand, insert, delete, sort. Skip list combines tree and linked list! So many choices!