From: Austin Ziegler <austin@...>
Date: 2003-10-03T22:22:05+09:00
Subject: Re: hash of hashes

On Fri, 3 Oct 2003 20:17:50 +0900, Paul Argentoff wrote:
> On Friday 03 October 2003 14:48, Emmanuel Touzery wrote:
>> not sure i understand your question. does this help you?
> Here is my example:
> irb(main):008:0> a = Hash.new(Hash.new(Array.new(2,0)))

This won't work because you're not doing what you're expecting. What
you're saying is that a nil key will return a particular Hash where
nil keys return a particular array.

This would be like saying:

  arry = Array.new(2, 0)
  hash = Hash.new(arry)
  a = Hash.new(hash)

  a['foo']        # => hash
  a['foo']['bar'] # => arry

What you want is:

  a = Hash.new do |h, k|
    h[k] = Hash.new do |h1, k1|
      h1[k1] = Array.new(2, 0)
    end
  end

This only works with 1.8 or bettter. Ruby does not auto-vivify
hashes or arrays like Perl. Note further that if you do something
like:

  a = Hash.new { |h, k| h[k] = :val; nil }

You will get a perhaps surprising -- but entirely correct -- result:
  
  a['foo']    # => nil
  a['foo']    # => :val

When you attempt to access a nonexistant hash key, the default value
is returned. In this case, it's nil. However, in the block you have
also set a side effect of setting the hash key provided to :val, so
the next time you access 'foo', it's no longer nonexistant.

-austin
--
austin ziegler    * austin@halostatue.ca * Toronto, ON, Canada
software designer * pragmatic programmer * 2003.10.03
                                         * 09.15.49