From: Stefano Crocco <stefano.crocco@...>
Date: 2008-05-28T04:15:50+09:00
Subject: Re: Can someone explain this syntax in new?

On Tuesday 27 May 2008, Robb Shecter wrote:
> I found this in an old post:
>
>
> "Here's a Hash that automatically creates new hashes for each key you
> try to access (specifically 1-level deep for a '2-dimensional' hash):
>
>   people = Hash.new{ |me,key| me[ key ] = {} }
>   people[ :gavin ][ :age ] = 34
>   people[ :gavin ][ :sex ] = :male
>   people[ :fido ][ :species ] = :dog
>   p people
>   #=> {:gavin=>{:age=>34, :sex=>:male}, :fido=>{:species=>:dog}}
>
>
>
> ...but there was no further explanation.  Could somebody explain the
> syntax in the first line?  Is this overwriting new()?
>
> Thanks,
> Robb

If you look at the ri documentation for Hash#new, you'll see that it can be 
used in three form: the first is Hash.new() which returns an hash with a 
default value of nil; the second is Hash.new(obj) which returns an hash with a 
default value of obj. This  means that whenever you try to access an element 
using a key which is not in the hash, the method will return obj. The third 
form of new takes a block. Whenever you try to access a nonexistent key, the 
block is called with two arguments: the hash itself and the key. In the block 
you can do whatever you want, including computing a value from the key and 
storing it in the hash.

In the code you posted, the block simply stores a new, empty hash under the 
key key of the hash.

I hope this helps

Stefano