From: Lyle Johnson <ljohnson@...>
Date: 2001-09-16T01:18:35+09:00
Subject: [ruby-talk:21210] Re: using my own objects as hash keys - a question

> Could someone explain to me why the following
> does not work as I expect:
>
<snip>

> hash=Hash.new
>
> hash[key1]=6
>
> puts hash[key1]  #=> 6
> # Ok, as expected
>
> puts hash[key2]  #=> nil
> # Why?!!  If key1.hash == key2.hash, why doesn't it work?
>
> # Same question here
> puts hash.has_key?(key1) #=> true
> puts hash.has_key?(key2) #=> false
>
>
> Have I missed something ?

Well, yes, I guess you've missed how hash tables work. A hash table is
usually described as a collection of "buckets", one for each hash value.
Whenever you add a new key to a hash table, the code will compute the hash
value for that key, find the corresponding bucket for that hash value, and
place a reference to the key in that bucket. When you look up that key in
the hash table (e.g. by calling has_key?) the code once again computes the
hash code for the key, looks in the corresponding bucket -- which may now
have several different keys, all with the same hash value -- and looks for
the requested key.

So even though key1 and key2 have the same hash code (6), you've only put
key1 in the hash table, and it's still a different instance than key2. If
you were to also add key2 to the hash table, they would both end up in the
same "bucket" and the hash table would include both keys.