A subtle difference between Ruby's Hash.fetch(:key, :default) vs. (Hash[:key] || :default)

defaults.rb

h = {
  'a' => :a_value,
  'b' => nil,
  'c' => false
}

h.fetch('a', :default_value) #=> :a_value
h.fetch('b', :default_value) #=> nil
h.fetch('c', :default_value) #=> false
h.fetch('d', :default_value) #=> :default_value

(h['a'] || :default_value) #=> :a_value
(h['b'] || :default_value) #=> :default_value
(h['c'] || :default_value) #=> :default_value
(h['d'] || :default_value) #=> :default_value

Good example

---

Is this difference only with MRI, what about JRuby, etc.? This company I'm working with deploys their ruby apps in production on Jruby, but we all develop locally on MRI, bitten me in the ass a few times. So just curious if you've discovered this difference between rubies/interpreter?

👍

For anyone who doesn't know, this is because Hash#fetch is interchangeable with hash.key?(key) ? hash[key] : default. I.e., #fetch checks for key existence, whereas || is testing for value falsiness. That's an extremely large distinction.

Note that {}[non_existent_key] still returns nil, but {}.fetch(non_existent_key) raises a KeyError.

Good info.

💯

Why use fetch if default value is not returned and nil can blow the application up? I'm now confused

Why use fetch if default value is not returned and nil can blow the application up? I'm now confused

fetch does return a default value, presuming you provide one. You can also provide a block, the return value of which, will be used as the "default value" in the case the key is missing from the hash.

You can get exactly same result if you do Hash.fetch(:key, ::default_value) || :default_value.