When you use YAML.dump
to serialize an object in Ruby, the class name is use as part of the Yaml tag so that the correct class can be used when loading the object. For example:
require 'yaml'
class Foo; end
puts YAML.dump Foo.new
produces
--- !ruby/object:Foo {}
When you use YAML.load
on that string, Psych knows what class to instantiate for the deserialized object.
If you try to call YAML.load
on a Yaml string that specifies a class that hasn’t been defined, then you will get the error:
require 'yaml'
# No Bar class has been defined
YAML.load '--- !ruby/object:Bar {}'
produces:
/Users/matt/.rvm/rubies/ruby-1.9.3-p327/lib/ruby/1.9.1/psych/visitors/to_ruby.rb:312:in `path2class': undefined class/module Bar (ArgumentError)
from /Users/matt/.rvm/rubies/ruby-1.9.3-p327/lib/ruby/1.9.1/psych/visitors/to_ruby.rb:312:in `resolve_class'
...
This is because Psych needs to create an instance of class Bar
, but doesn’t have the definition of the class available. This explains why adding require 'whatever'
before loading the Yaml works – now Ruby has the definition of the class loaded and so can create an instance of it (note that there is no definitive link between class name and file name in Ruby, it’s just convention).
The solution therefore is to make sure that when you’re loading any Ruby objects from Yaml you have already required any files that may contain definitions of any classes potentially in that Yaml.