The way Unix processes work is rather interesting, and beyond the scope of this blog, but worth the investigation. I will cover the basics.

Every process running in the user space has a parent process. If you follow a processes ancestry you will arrive at the init process (process id of 1). The init process is, itself, a child of the sched or swapper process (process id 0) which is a special process run by the kernel.

The way that an application is loaded in Unix is that the current process is forked. The subprocess inherits all of its parent’s globals:

  • environment variables
  • open file descriptors
  • data/memory/address space (look up copy-on-write)
  • code/text

The child process get it’s own execution stack

After forking, an exec call is made and replaces the current code/text and data with a desired application’s code and data, etc. This is important information for any coder to know, so know processes, forking and exec inside and out.

Let’s take a step back though and look at the Ruby forking big picture.

Let Us Code

In Ruby, you can create a child a process using Process#fork method.

The fork method returns twice, once in the parent process and once in the subprocess, also called the child process. The first value returned is an integer, the child process id, which is returned from the parent process. The second value returned is nil and returns from the child process.

$ ruby -e "p fork"
# => 12345  # This is returned from the parent process
# => nil    # This is returned from the child process

Now that we have that out of the way, let’s create a demo script. For this demo, open a second terminal.

# myfork.rb
puts "My pid is #$$"

if child_id = fork 
    puts "I am a parent, my id is #$$ and my child is #{child_id}"
    puts "I am a child, my id is #$$"


In terminal 1, run the program and let it block:

$ ruby myfork.rb
My pid is 21830
I am a parent, my id is 21830 and my child is 21859
I am a child, my id is 21859

In terminal 2, list the processes:

$ ps - | grep ruby
pricees  21830 20005  0 11:36 pts/6    00:00:00 ruby myfork.rb
pricees  21859 21830  0 11:36 pts/6    00:00:00 ruby myfork.rb

Press a key in the terminal 1 to exit the program.

Knowing that the parent process returns a truthy-value (integer) and that the child process returns a falsy-value (nil), we can use the fork method as a predicate to branch between two different execution paths.

# File: fork_that_thang.rb

if (child = fork)
    puts "Waiting for child (#{child}) to count to 5"

    # Wait for child process to exit
    (1..5).each do |i|
        sleep 1
        puts i

Waiting for child (6495) to count to 5

Sweeeeeeet, right?!


Forking is one of the fundamental system calls to make all applications run in the user space. A developer should understand this system call inside-and-out. Leveraging fork in Ruby will allow the developer to create applications of non-trivial complexity.

Go forth and conquer!

blog comments powered by Disqus


10 September 2015