Rinda-based Resque clone

require 'rinda/tuplespace'

# Workers will be initialized a Rinda::TupleSpace
# (which we can use as a stand-in for Redis

class MapReduceWorker
  def initialize(tuplespace)
    @ts = tuplespace
    @name = "MapReduceJob"
  end

  def work
    return if @ts.read([@name, nil])

    job = @ts.take([@name, nil]).last
    puts "Performing a MapReduceJob!"
    result = job[:data].inject(&job[:method])
    puts "computed result: #{result}"
  end
end

# bin/caribou is responsible for initializing workers,managing the TupleSpace
URI = "druby://localhost:7777"
ts = Rinda::TupleSpace.new

# initialize all the workers, three steps:
# 1. create a worker
m = MapReduceWorker.new(ts)

# 2. create an observer which will fire when
# a compatible item is added to the TupleSpace
observer = ts.notify 'write', ['MapReduceJob', nil]

# 3. kick off a thread with a block to call worker.work, when something published to TupleSpace
Thread.start do
  observer.each { |t| m.work }
end


DRb.start_service(URI, ts)
DRb.thread.join

# then another process has to write Jobs to the tuplespace
#
Caribou.connect("druby://localhost:7777")

def some_controller
  #...
  Caribou.enqueue([MapReduceWorker, {data: [1,2,3,4,5], method: :+ }])
end