Hello everyone,
I'm using the rails-settings gem, and I'm trying to understand how you add functions to ActiveRecord classes (I'm building my own library for card games), and I noticed that this gem uses one of the Meta-programming techniques to add the function to the ActiveRecord::Base class (I'm far from Meta-programming master in ruby, but I'm trying to learn it)
module RailsSettings
class Railtie < Rails::Railtie
initializer 'rails_settings.initialize', :after => :after_initialize do
Railtie.extend_active_record
end
end
class Railtie
def self.extend_active_record
ActiveRecord::Base.class_eval do
def self.has_settings
class_eval do
def settings
RailsSettings::ScopedSettings.for_thing(self)
end
scope :with_settings, :joins => "JOIN settings ON (settings.thing_id = #{self.table_name}.#{self.primary_key} AND
settings.thing_type = '#{self.base_class.name}')",
:select => "DISTINCT #{self.table_name}.*"
scope :with_settings_for, lambda { |var| { :joins => "JOIN settings ON (settings.thing_id = #{self.table_name}.#{self.primary_key} AND
settings.thing_type = '#{self.base_class.name}') AND
settings.var = '#{var}'" } }
scope :without_settings, :joins => "LEFT JOIN settings ON (settings.thing_id = #{self.table_name}.#{self.primary_key} AND
settings.thing_type = '#{self.base_class.name}')",
:conditions => 'settings.id IS NULL'
scope :without_settings_for, lambda { |var| { :joins => "LEFT JOIN settings ON (settings.thing_id = #{self.table_name}.#{self.primary_key} AND
settings.thing_type = '#{self.base_class.name}') AND
settings.var = '#{var}'",
:conditions => 'settings.id IS NULL' } }
end
end
end
end
end
end
What I don't understand is why he uses class_eval on ActiveRecord::Base, wasn't it easier if he just open the ActiveRecord::Base class and define the functions? Specially that there's nothing dynamic in the block (What I mean by dynamic is when you do class_eval or instance_eval on a string containing variables)
something like this:
module ActiveRecord
class Base
def self.has_settings
class_eval do
def settings
RailsSettings::ScopedSettings.for_thing(self)
end
scope :with_settings, :joins => "JOIN settings ON (settings.thing_id = #{self.table_name}.#{self.primary_key} AND
settings.thing_type = '#{self.base_class.name}')",
:select => "DISTINCT #{self.table_name}.*"
scope :with_settings_for, lambda { |var| { :joins => "JOIN settings ON (settings.thing_id = #{self.table_name}.#{self.primary_key} AND
settings.thing_type = '#{self.base_class.name}') AND
settings.var = '#{var}'" } }
scope :without_settings, :joins => "LEFT JOIN settings ON (settings.thing_id = #{self.table_name}.#{self.primary_key} AND
settings.thing_type = '#{self.base_class.name}')",
:conditions => 'settings.id IS NULL'
scope :without_settings_for, lambda { |var| { :joins => "LEFT JOIN settings ON (settings.thing_id = #{self.table_name}.#{self.primary_key} AND
settings.thing_type = '#{self.base_class.name}') AND
settings.var = '#{var}'",
:conditions => 'settings.id IS NULL' } }
end
end
end
end
I understand the second class_eval (before the def settings) is to define functions on the fly on every class that 'has_settings' right ? Same question here, I think he could use "def self.settings" instead of "class_eval.... def settings", no ?