RoR - different object_id inside after_initialize callback after obj.reload - mysql

Question

I've written a method for my project which extends ActiveRecord models behaviour, I've stripped out most of it, consider the following code:
class ActiveRecord::Base
def self.has_translations
after_initialize :clear_translations_cache
def clear_translations_cache
binding.pry
#_translations = {}
end
end
end
Basically, I want the #_translations instance variable to get cleared when I .reload the instance from the database, but for some reason, after fetching an existing object from the database, executing a method which populates #_translations, and then executing object.reload, #_translations still contains the same data.
I know for sure that the callback gets executed when first fetching the object from database and when calling .reload. I used binding.pry to halt execution inside the callback method, but for some reason self.object_id inside .reload has a different object_id than my original object, and therefore #_translations in the original object doesn't get cleared.
Attached is the console output:
1.9.3p194 :008 > s = TranslatedItem.first
76: def clear_translations_cache
=> 77: #_translations = {}
78: end
[1] pry(#<TranslatedItem>)> self.class
=> TranslatedItem(id: integer, created_at: datetime, updated_at: datetime)
[2] pry(#<TranslatedItem>)> self.object_id
=> 70254243993580
[3] pry(#<TranslatedItem>)> exit
1.9.3p194 :009 > s.object_id
=> 70254243993580
1.9.3p194 :010 > s.reload
76: def clear_translations_cache
=> 77: #_translations = {}
78: end
[1] pry(#<ServiceLevel>)> self.class
=> TranslatedItem(id: integer, created_at: datetime, updated_at: datetime)
[2] pry(#<TranslatedItem>)> self.object_id
=> 70254259259120

Answer 1

I'm guessing the behavior you're seeing is related to how ActiveRecord reload works:
fresh_object = self.class.unscoped { self.class.find(self.id, options) }
#attributes.update(fresh_object.instance_variable_get('#attributes'))
You'll notice that it is creating a fresh object by finding it from the database, which explains why you are seeing two different object_id values in your callback method. The object that is initialized during the reload is only used for its attributes and then goes out of scope.
It's not clear from your question whether you were just curious why it behaved this way or if you're looking for an alternative way to do it.
Update:
You've got a few options if you're just looking for a way to clear the instance variable when you reload the model.
1) Add your own reload method that you can explicitly call:
class ActiveRecord::Base
def reload_everything
reload
#_translations = {}
end
end
object.reload_everything
2) Change the behavior of reload
module ReloadTranslations
def reload(*args)
super
#_translations = {}
end
end
ActiveRecord::Base.send(:include, ReloadTranslations)