Futurist

Background

Goals of Futurist

Provide a mechanism to transparently execute code in parallel.

On Futures, Promises, and Resolution

A future is a proxy for a result that is initially unknown, usually because the computation of its value is not yet complete.

The terms Future and Promise are colloquially used as interchangably, however their formal definitions describe important, distinct concepts. Formally, a Future is something which holds a value whereas a Promise is the funciton, or means by which the future's value is set; note that a promise may be, and is often, executed asyncrhonously.

"A future is a read-only placeholder view of a variable. A promise is a writable, single assignment container which sets the value of the future." ¹

When a promise computes and sets the value of its future, it is said to resolve the value of the future.

Features of Futurist

Eager promise resolution. Promise resolution begins during Future initialization. Futurist resolves its promises in a forked process. This allows promises to be resolved in parallel.
Blocking synchronization semantics. If the value of a future is requested before its promise is resolved, the call to value blocks until the promise is resolved. This removes the need for special-case error handling and allows the client code to be ignorant of synchronization code and interact with the future in an imperative manner.

When to use Futurist

A set of tasks has inherent latency for each task. For example, making a set of HTTP requests.
Task execution benefits from memory isolation provided by process forking.

When NOT to use Futurist

If the runtime operating system does not support process forking. Specifically, "...fork(2) is not available on some platforms like Windows and NetBSD 4." ²
If the promise needs to manipulate the process hierarchy. For example, if the promise makes use of Kernel#exec; Kernel#exec forks a child process and replaces the current process with the newly forked process. This is problematic in the promise as it replaces the value resolving process.
The set of active futures at any given time is greater than the maximum number of file descriptors a user is allowed to open on the file system. This can be mitigated by processing the set of tasks in batches.
The overhead of process forking outweighs the cost of the latency for the set of tasks to be performed.

Installation

Directly via RubyGems

$ gem install futurist

then

require 'futurist'

or with Bundler

add this line to your application's Gemfile:

gem 'futurist'

And then execute:

$ bundle

Usage

Create a future

The call to Futurist::Future#value will block until the result of executing the block is available. If an exception occurs during the block's execution, the call to future.value will reraise the exception.

future = Futurist::Future.new { 3 + 2 }
future.value # blocks until value is available
=> 5

Development

After checking out the repo, run bin/setup to install dependencies. Then, run rake rspec to run the tests. You can also run bin/console for an interactive prompt that will allow you to experiment.

To install this gem onto your local machine, run bundle exec rake install. To release a new version, update the version number in version.rb, and then run bundle exec rake release, which will create a git tag for the version, push git commits and tags, and push the .gem file to rubygems.org.

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/indiebrain/futurist.

indiebrain/futurist