Hash#shift using default values

How the way you define your hash default value can generate side effects in your program?

The 3 main ways to declare a default value to a Hash are:

• by using the Hash.new('default value') method
• by using the Hash#default= method
• by using the Hash.new { |h, k| h[k] = 'default value' }

Each of these 3 methods produces a different behavior using the Hash#shift method.

Hash#shift

The hash#shift method removes and returns a pair key/value of a hash

Here, we can see that the initial hash contains a pair foo: :bar .

After the call to hash.shift , the hash is empty and the pair [:foo, :bar] is returned and stored in the pair variable.

Then if we call the hash.shift method on an empty hash nil is returned.

Now that we’re more familiar with the Hash#shift method, let’s see how it behaves when the hash defines a default value.

Hash#default=

Here, we define an empty hash . Then we set the 'val' default value for this hash .

Finally, when we call the hash.shift method we notice that our default value is returned as the hash is empty.

Hash.new(default_value)

Here we define an empty hash using the Hash.new with an argument as default value.

When we call the hash.shift method, we notice the exact same behavior as a hash that sets default value using the Hash#default= method.

Hash.new {}

Here we define an empty hash using the Hash.new with a block that defines the default value.

Then we call the hash.shift method that returns the hash default value.

But a second call to hash.shift returns a pair with nil as key and our default value as value.

Weird..

Not at all. Let’s detail what happens here in the following section.

blocks and Hash#shift

Let’s add few lines to our previous example to understand what’s happens step-by-step

Here, we display the content of our hash after a call to hash.shift and we notice that it contains a nil => 'val' pair.

Then we call hash.shift for the second time and logically this pair is returned.

So, why this pair is inserted in our hash after the first call to hash.shift ?

The answer is simple: because the block passed to Hash.new is executed each time that you code tries to access the default value of our hash .

Indeed, the call to hash.shift forces our hash to execute the block passed to Hash.new as its return value is used as default value of hash — when hash is empty.

But the subtlety here is that k in the |h, k| block arguments is nil .

So, the block is processed as following:

hash[nil] = ‘val’
hash#[]=
'val'

This is why the first call to hash.shift return 'val' and sets the pair { nil => 'val' } in our hash .

So a second call to hash.shift logically returns [nil, 'val'] and not the default value as the hash is not empty at this stage.

Conclusion

When using Hash#shift or any methods that interact with the default value of the hash, be aware if you use a block as default value of the hash.

Indeed, any assignations within this block can generate side effects that can roughly alter the behavior of your app.

Voilà