Will Clarke

Real Defaults With Hashes


Beware of setting Hash#default!

Unless you know the pitfalls of Hash#default, you should tread carefully…

You can set default values for hashes a number of ways. Eg:

  1. Hash.new(default_value)
  2. Hash.new.default = default_value

Hash.new(0) usually works as expected, but Hash.new([]) can be tricky.

If you wanted to append any values to a hash instantiated like this, there will be problems; the array is not just a default array, but it’s also a shared array.

An example:

h = Hash.new([])
h['fish'] << 'carp'
puts h  # => nil.

What’s happening here is that we’ve edited the default value of the hash. We haven’t actually assigned the default value to our lovely ‘fish’. To actually assign, we need to do:

h['fish'] <<= 'carp'

Note the = sign. However, this is not perfect:

puts h  # => {fish: ['carp', 'carp']}

Our ‘default’ value how has two carp in it. If we were to add another key-value pair, this would get even messier:

h['dog'] <<= 'poodle'
puts h  # => {fish: ['carp', 'carp', 'poodle'], dog:['carp', 'carp', 'poodle']}

A disaster!

Our carp and Poodle have now mixed. Hopefully it makes sense what’s happening; we’re altering the (single) default array, which points to all our default values. How do we resolve the situation?

Blocks to the rescue!

You can pass in a block when initializing the Hash.

Hash.new {|hash, key| ... rest of block... }

To create a default hash which hash a unique default for each key/value pair, we can then do this:

Hash.new {|hash, key| hash[key] = [] }

Since this block is run each time we assign a new key, the default value will always be [].

See also Hash#default_proc.

The benefit of using blocks & procs is that we’re able to write clever pieces of code.. and could make our default value do anything we wanted.


ruby tips