propaganda

The propaganda library is a Clojure implementation of the propagator computational model described in The Art of the Propagator. The aim of the library is to be easy to use and reason about, yet extensible.

Two different strategies have been implemented: one using the Clojure STM for handling the propagation of values in the system; and one representing the system as an immutable value, without the aid of any underlying transactional model. The latter approach makes it possible to use propagators from javascript, and is the biggest contribution from this project.

There is a Clojars release containing an implementation that works in both Clojure and ClojureScript.

The leiningen dependency is

[propaganda "0.2.0"]

Tutorial

Here follows a short tutorial. For more in depth information, please consult the following sources:

• The 2013 Clojure/conj talk Propagators in Clojure, which gives a broad overview of what propagators are, and how they are implemented in this library.

• The 2014 London Clojurians meetup talk "Propaganda! A declarative programming model in Clojure" is available here as a SkillsCast. The examples and the slides are available here at github.

• Getting hot with propagators and Taking propagators to the next level - two tutorials that dive a bit deeper in explaining how the propaganda library works.

• The gh-pages branch of this project contains a brief explanation of the basics of the propagator computational model. The page can be visited here.

• STM vs. System propagation illustrates the differences between the STM and system approach to propagation using the building height problem.

• Extending propaganda with set values shows how to extend propaganda with support for sets.

• The Art of the Propagator contains a more in-depth explanation of how propagation can be implemented.

To use the propaganda library, you need to define a merge, function, create cells and set up propagators. The merge function is invoked when a propagator attempts to store a new value in a cell. The merge function is invoked with the current value and the new value, and must return either a new value which will be stored in the cell, or a Conflict object.

In this short example we just use the default merger function, we define the square and square-root propagator and set up relations beween simple cells.

(use 'propaganda.stm)
(use 'propaganda.values)

default-merge will give us a merger that will merge nothing with anything, but will enforce that anything else that is attempted to be merged will return a contradiction

(def my-merge
  (default-merge))

nothing can be merged with nothing and will return nothing

(my-merge nothing nothing)
;; => :propaganda.values/nothing

Anything else will be the result of the merge

(my-merge nothing 1)
;; => 1
(my-merge 2 nothing)
;; => 2
(my-merge 1 1)
;; => 1

... unless it gives rise to a contradiction

(my-merge 1 2)
;; => #propaganda.core.Contradiction{:reason "1 != 2"}

The function->propagator-constructor can be used for setting up simple one way relations

(def squarer
  (function->propagator-constructor
   (fn [val] (* val val))))

(def sqrter
  (function->propagator-constructor
   (fn [val] (Math/sqrt val))))

... which can be extended to go both ways

(defn quadratic
  [x x-squared]
  (squarer x x-squared)
  (sqrter  x-squared x))

We can now construct cells and set up the quadratic relations to read the squared of a number in our system:

(let [x (make-cell)
      x-squared (make-cell)]
  (binding [*merge* my-merge]
    (quadratic x x-squared)
    (add-content x 10.0)
    (get-content x-squared)))
;; => 100.0

Or the square-root, depending on the input from the user

(let [y (make-cell)
      y-squared (make-cell)]
  (binding [*merge* my-merge]
    (quadratic y y-squared)
    (add-content y-squared 1764.0)
    (get-content y)))
;; => 42.0

We will be warned of any inconsistencies in our system when adding content

(let [z (make-cell)
      z-squared (make-cell)]
  (binding [*merge* my-merge]
    (quadratic z z-squared)
    (add-content z 10.0)
    (add-content z-squared 123.0)))
;; Exception: Inconsistency: 100.0 != 123.0

Motivation

The objective of this project is to create an extinsible propagator library for Clojure. Propagators define a declarative computational model. They are described in the article The Art of the Propagator.

Along with the library itself, the project should supply documentation of the API, good examples and tutorials.

I have not previously worked with propagtors, so this will also be an exploration for me.

Thanks to

Ragnar Dahlén and Kasper Langer for feedback on the library.