Skip to main content

Visitor Pattern Deprecated: First-Class Messages Are All You Need!

In previous posts [1,2], I argued that the visitor pattern was a verbose, object-oriented equivalent of a pattern matching function. The verbosity stems from the need to add the dispatching code to each data class in the hierarchy, and because the encapsulation inherent to objects is awkward when dealing with pure data.

A single addition to an OOP language could completely do away with the need for the dispatching code and make OO pattern matching simple and concise: first-class messages (FCM). By this I mean, messages sent to an object are themselves 'objects' that can be passed around as parameters.

To recap, functional languages reify the structure of data (algebraic data types), and they abstract operations (functions). OOP languages reify operations (interfaces), but they abstract the structure of data (encapsulation). They are duals of one another.

All the Exp data classes created in [1,2] shouldn't be classes at all, they should be messages that are sent to the IVisitor object implementing the pattern matching:

data Exp = App(e,e) | Int(i) | ...

IVisitor iv = ...

//the expression: 1 + 2
Exp e = App(Plus(Int(1), Int(2)));

//'e' is now the "App" operation placed in a variable
iv.e  //send "App" to visitor

FCM removes the inefficient double-dispatch inherent to the visitor pattern, while retaining object encapsulation for when you really need it; the best of both worlds.

Note also how the above message declaration looks a great deal like a variant in OCaml? This is because first-class messages are variants, and pattern-matching functions are objects. I'm actually implementing the DV language in that paper, first as an interpreter, then hopefully as a compiler for .NET. To be actually useful as a real language, DV will require some extensions though, so stay tuned. :-)

[Edit: made some clarifications to avoid confusion]
Labels:

Comments

Post a Comment

Popular posts from this blog

Blue-eyed Islander Puzzle - an analysis

Many people find themselves stumped by the so-called Blue-Eyed Islanders puzzle . There is also much controversy over its supposed solution. I'm going to analyze the problem and the solution, and in the process, explain why the solution works. To begin, let's modify the problem slightly and say that there's only 1 blue-eyed islander. When the foreigner makes his pronouncement, the blue-eyed islander looks around and sees no other blue eyes, and being logical, correctly deduces that his own eyes must be blue in order for the foreigner's statement to make sense. The lone blue-eyed islander thus commits suicide the following day at noon. Now comes the tricky part, and the source of much confusion. Let's say there are 2 blue-eyed islanders, Mort and Bob. When the foreigner makes his pronouncement, Mort and Bob look around and see only each other. Mort and Bob thus both temporarily assume that the other will commit suicide the following day at noon. Imagine their chagrin...
40 comments
Read more

Easy Reverse Mode Automatic Differentiation in C#

Continuing from my last post on implementing forward-mode automatic differentiation (AD) using C# operator overloading , this is just a quick follow-up showing how easy reverse mode is to achieve, and why it's important. Why Reverse Mode Automatic Differentiation? As explained in the last post, the vector representation of forward-mode AD can compute the derivatives of all parameter simultaneously, but it does so with considerable space cost: each operation creates a vector computing the derivative of each parameter. So N parameters with M operations would allocation O(N*M) space. It turns out, this is unnecessary! Reverse mode AD allocates only O(N+M) space to compute the derivatives of N parameters across M operations. In general, forward mode AD is best suited to differentiating functions of type: R → R N That is, functions of 1 parameter that compute multiple outputs. Reverse mode AD is suited to the dual scenario: R N → R That is, functions of many parameters t...
18 comments
Read more

The Chicken and the Egg - Redux

There is still considerable skepticism regarding my conclusion that the chicken comes first . Many of the objections are simply due to different interpretations of the question, interpretations which I consider unfaithful to the original purpose of the chicken-egg dilemma . Fundamentally, this question is supposed to represent a causality dilemma. When there is a causal dependency between any two objects, A and B, we can ask ourselves, "which came first, A or B?" An object C could have caused B, and thus triggered the recursive relation C→B→A→B→A... Of course, it could just as easily have been A that was caused first. To properly answer this question, we must reduce the abstract objects A and B to concrete objects and apply our scientific knowledge to ground the recursion. Using chickens and eggs as our objects, we have to precisely define what chickens and what eggs to consider. The question "which came first, chickens or any type of egg?" is not a causal dilemma a...
29 comments
Read more