GOOGLE?????!!!! jk.

A hash table is typically an array with N buckets (positions/slots/entries) in it, often a power of two. Each bucket contains room for one or more items. The hash value happens to be the index into that array. Why is it called a bucket? I'm not sure.

A bucket is typically a watertight, vertical cylinder or truncated cone, with an open top and a flat bottom, usually attached to a semicircular carrying handle called the bail. In common usage, the terms "bucket" and "pail" are often used interchangeably. As a shipping container, a pail is a package with a sealed top or lid.

There are many types of buckets;

• a water bucket is used to carry water

• Household and garden uses are often for carrying liquids and granular products

• Elaborate ceremonial or ritual buckets in bronze, ivory or other materials are fond in several ancient or medieval cultures and are known by the Latin for bucket, situla.

• Large scoops or buckets are attached to loader and telehandler for agricultural and earthmoving purposes.

• A lunch box is often called a lunch pail

• Buckets can be reused as seats, tool caddies, hydroponic gardens, chamberpots, "street" drums, livestock feeders.

Most of the data structure books I have ( actually all) have a chapter on hashing and not once do they mention buckets, but after searching google I found some examples.

Yes, his coding style must consist of typing english.

Chess programs use hash tables to store search results, with the hash key generated from the board position. Typically, the table is stored as a plain array of 2^N elements and the index is calculated from the lowest N bits of the hash key.
The number of positions searched is very much larger than the number of elements in the table, so you need multiple buckets per entry (in practice, one looks at the next k entries, say 4 (you really would like to stay within the same cache line)) and some replacement strategy to get rid of older entries.

Not sure if that's at all helpful for what you want to do...

I suppose. But I can be lazy and don't want to implement a dynamically resizing hash table, and still want tens/hundreds of millions of entries. The resize operation has a rather large amount of overhead on large tables, you have to iterate over every single item and rehash it into the new table. kinda sucky.

Boost's unordered_map uses buckets. See here.

std::unordered_map<> (new in C++11) also uses buckets (it's nearly identical to the boost implementation). Before C++11, no other standard STL container uses buckets.

You don't iterate through buckets, you iterate through elements. boost::unordered_map<> supports forward iterations like any other STL container.

No specified support for backwards iterations though (it may or may not work on your implementation).

I want to make a clone of a boost unordered_map. I'm just doing it to learn the fundamentals of hash tables but I really liked that one so I picked that to learn from.

I did find chapter 23 in Professional C++ full of useful information.