From: Martin DeMello <martindemello@...>
Date: 2005-11-10T01:57:13+09:00
Subject: Re: Nested hash constructor confusion

Ben Armstrong <BArmstrong@dymaxion.ca> wrote:
> Can anyone explain the following?  Our "intuitive" solution (h1) to
> creating a constructor for a hash nested within a hash did not work.  It
> was only after attempting a workaround (h3) and finding that it
> performed very poorly that we Googled (h2) a correct solution.  But it
> is not obvious to us why this more complex constructor is required.

Hash.new, when passed a block, stores the block as a sort of
"key_missing" callback, passing it the hash itself and the key. The
callback doesn't modify the hash, though, it just returns a virtual
value. However, because the block does have access to the hash, it can
call a method on that hash that updates it, hence the second
constructor you tried.

Compare:

irb(main):001:0>  h = Hash.new {|h, k| 2*k}
=> {}
irb(main):002:0> h[1]
=> 2
irb(main):003:0> h[2]
=> 4
irb(main):004:0> h
=> {}
irb(main):005:0> h1 = Hash.new {|h, k| h[k] = 2*k}
=> {}
irb(main):006:0> h1[1]
=> 2
irb(main):007:0> h1[2]
=> 4
irb(main):008:0> h1
=> {1=>2, 2=>4}

And even something more complicated like:

irb(main):009:0> h2  = Hash.new {|h, k| h[k] = 2*k; "missing, filling it
in"}
=> {}
irb(main):010:0> h2[1]
=> "missing, filling it in"
irb(main):011:0> h2[1]
=> 2

martin

martin