Project

dunder

0.01
No commit activity in last 3 years
No release in over 3 years
For tasks that can be started early and evaluated late. Typically one might want start multiple heavy tasks concurrent. This is already solvable with threads or the [reactor-pattern](http://rubyeventmachine.com/) but setting this up could be cumbersome or require direct interactions with threads ex. Dunder is a simple way of abstracting this: you simply pass a block to Dunder.load with the expected class as the argument
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
 Dependencies

Development

>= 3.0.3
~> 1.0.0
~> 1.5.2
>= 0
 Project Readme

Dunder

For tasks that can be started early and evaluated late.

A simple way of doing heavy work in a background process and blocking until done when you really need the object.

Preloading using the proxy pattern Heavily inspired by Adam Sandersons post

Introduction

To increase performance typically one might want start multiple heavy tasks concurrent. This is already solvable with threads or the reactor-pattern but setting this up could be cumbersome or require direct interactions with threads etc.

What inspired me was the ability to run concurrent database queries within a single request in rails, please read more in the section below.

How you could lazy load something today in ruby 1.9

foo = "foo"
bar = "bar"
t = Thread.start {
	sleep 1
	foo + bar
}
# Other code
foobar = t.value

The Thread.start call would not block and execution would continue and when you need the value you could ask t.value for it

Dunder is a simple way of abstracting this but does infact use threads behind the scenes: you simply pass a block to Dunder.lazy_load When later accessing the returned object, lets say: lazy_object will block until the thread is done and has returned or if the thread is done returns the value.

Dunder will only be happy under 1.9.* because how blocks changed have changed. There also some caveats that you should read about below

Usage

lazy_object = Dunder.lazy_load {
	# heavy stuff
	value
}
# Later on access lazy_object
puts lazy_object
puts lazy_object.class # => value.class

or through chaining with dunder_load which works for both objects and classes

lazy_sorted_array = array.dunder_load.sort

# With arguments and block
lazy_obj = obj.dunder_load.do_something_heavy(a,b,c) {
	#maybe something other heavy here
}

Parallel example

lazy_foo = Dunder.lazy_load {
	# Simulate heavy work
	sleep 2
	"foo" 
}

lazy_bar = Dunder.lazy_load {
	# Simulate heavy work
	sleep 2
	"bar" 
}

# Do something other heavy

puts lazy_bar # => "bar"
puts lazy_bar.class # => String
puts lazy_foo # => "foo"
puts lazy_foo.class # => String
# Will finish after 2 seconds

worth mentioning is that if you access the variable in someway before that it will block earlier ex

lazy_array = Dunder.lazy_load do
	sleep 1
	[1,2,3]
end
puts lazy_array.length # <- will block here until the above sleep is done
sleep 1 # other heavy stuff
puts lazy_array # <- will be printed after 2 seconds

changing the order of the statements will fix this though

lazy_array = Dunder.lazy_load do
	sleep 1
	[1,2,3]
end
sleep 1 # other heavy stuff
puts lazy_array.length # <- will block here until the above sleep in the block is done
puts lazy_array # <- will be printed after 1 second

WARNING "if-it-quacks-like-a-duck-walks-like-a-duck"

  • Don't return symbols
  • And for normal objects be careful with comparing

The reason for this is that the implementation uses the delegation.rb in ruby which makes objects life tricky. Even though the object return quacks like a object it will not always walk in a straight line. Ex

o = Object.new
res = Dunder.lazy_load { o }
res == o # => true
o == res # => false 
o == res._thread.value # => true

But Array,String,Fixnum,Hash etc work fine.

If you want to be sure that nothing fishy is going on please use ._thread.value

Groups

So now you might be wondering what would happen if we lazy load more than 10000 objects through some intense calculations. Well our performance would decrease because of the context switching, it would actually be better if we only ran a limited number of lazy loads at any one point.

Dunder::Group has been specifically designed to solve this problem.

For our contrived example note that this example and dunder requires a ruby version of at least 1.9.* . Lets say we have list of tens of thousands of urls that we want to visit and measure the sum content length of all the websites

require 'open-uri'
list = ["http://google.com","http://yahoo.com", .... ]
g = Dunder::Group.new(100)
results = list.map do |u|
  g.lazy_load { open(u) } 
  # or dunder_load(g).open(u)
end

sum = 0
results.each do |r|
  sum += r.length
end
puts sum

Note that you could use groups by itself and pass blocks

g = Dunder::Group.new(4)
t = g.start_thread {
  # things to do here
}

Much depending on what you are doing you will want to pick a higher or lower number. If your task is CPU-bound then around the number of cores on your computer should be optimal, if your task is IO bound which is true for most of my use cases then experimenting is key.

Rails

# Will not block
@lazy_posts = Post.dunder_load.all

@lazy_user = User.dunder_load.first

and then later in views

<%= @user.name %> <-  will block until the user have been loaded
<%= @lazyposts.each do %> <- will block until the posts have been loaded
...
<% end %

Be careful not to use the mysql gem which blocks the whole universe on every call. Please use the mysql2 which is the standard adapter for rails since 3.0, also the pg gem works fine.

For a sample application using mysql checkout this

Install

gem install dunder

(The MIT License)

Copyright (c) 2011 Erik Fonselius

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.