Question

I have a case where I’m trying to leverage a database configuration file used by Padrino for a custom worker I’m writing without having to load the Padrino environment and without having to modify the existing database configuration. In short, I want to write my code to work with the existing database configuration code rather than modifying the existing configuration.

The database configuration file in question looks like the following:

# This is just here for right now so I can test
# to see if logger is set to anything in DB config
puts "Logger in DB config: #{logger}"

ActiveRecord::Base.configurations[:development] = {
  :adapter => 'sqlite3',
  :database => Padrino.root('db', "app_development.db") }
ActiveRecord::Base.logger = logger
ActiveRecord::Base.include_root_in_json = true
ActiveRecord::Base.store_full_sti_class = true
ActiveSupport.use_standard_json_time_format = true
ActiveSupport.escape_html_entities_in_json = false
ActiveRecord::Base.establish_connection(ActiveRecord::Base.configurations[Padrino.env])

My custom worker looks like the following:

ROOT = File.expand_path(File.dirname(__FILE__))
$LOAD_PATH.unshift(ROOT)

require 'active_record'
require 'logger'

# Mock Padrino module with methods used by database config file
module Padrino
  def self.root(*args)
    return File.expand_path(File.join("#{ROOT}/..", *args))
end

  def self.env
    return :development
  end
end

module Worker
  class << self
    attr_accessor :logger

    def init
      @logger       = Logger.new(STDOUT)
      @logger.level = Logger::DEBUG

      require "#{ROOT}/../config/database"
    end
  end
end

Worker.init

When I require the database configuration file in my worker I get an error saying the following:

/home/user/devel/app/config/database.rb:3:in `<top (required)>': undefined local variable or method `logger' for main:Object (NameError)
from /home/user/.rvm/rubies/ruby-1.9.2-p290/lib/ruby/site_ruby/1.9.1/rubygems/custom_require.rb:55:in `require'
from /home/user/.rvm/rubies/ruby-1.9.2-p290/lib/ruby/site_ruby/1.9.1/rubygems/custom_require.rb:55:in `require'
from worker.rb:26:in `init'
from worker.rb:31:in `<main>'

Given this, I modified the Worker#init function to be the following:

def init
  @logger       = Logger.new(STDOUT)
  @logger.level = Logger::DEBUG

  Object.send(:define_method, :logger, lambda { @logger })

  puts "Logger in Worker module: #{logger}"

  require "#{ROOT}/../config/database"
end

This change resulted in the following output:

Logger in Worker module: #<Logger:0x9505088>
Logger in DB config:

I take this to mean the @logger is not in scope in the database configuration file, even though I’m using a lambda when I define the logger method on Object.

Oddly enough, if I pull the Object.send line of code out of the Worker#init function and instead call it right before I call Worker.init like below, I get the following result.

Object.send(:define_method, :logger, lambda { Worker.logger })
Worker.init

results in

Logger in Worker module: #<Logger:0x81bc8b4>
Logger in DB config: #<Logger:0x81bc8b4>

Can someone explain to me why if I make the Object.send call inside the Worker#init function it doesn’t work like it does if I make it outside the Worker module?

Answer 1

In your first example, you try to use undefined variable ‘logger’, as you can see from error message. Try to use the logger feature this way:

ActiveRecord::Base.logger = Logger.new(STDOUT)