Andre Staltz的这篇文章是迄今为止我所见过的最好、最清楚的解释。

以下是文章中的一些引述:

响应式编程是使用异步数据流进行编程。 最重要的是，你会得到一个神奇的功能工具箱来组合、创建和过滤任何这些流。

下面是文章中精彩图表的一个例子:

Andre Staltz的这篇文章是迄今为止我所见过的最好、最清楚的解释。

以下是文章中的一些引述:

响应式编程是使用异步数据流进行编程。 最重要的是，你会得到一个神奇的功能工具箱来组合、创建和过滤任何这些流。

下面是文章中精彩图表的一个例子:

如果你想感受一下FRP，你可以从1998年的Fran教程开始，它有动画插图。对于论文，从函数反应动画开始，然后在我的主页上的出版物链接和Haskell wiki上的FRP链接上跟踪链接。

就我个人而言，我喜欢在讨论如何实施FRP之前思考它意味着什么。 (没有规范的代码是没有问题的答案，因此“甚至没有错”。) 因此，我没有像Thomas K在另一个答案(图、节点、边、触发、执行等)中那样用表示/实现术语描述FRP。 有许多可能的实现风格，但没有一种实现说明FRP是什么。

I do resonate with Laurence G's simple description that FRP is about "datatypes that represent a value 'over time' ". Conventional imperative programming captures these dynamic values only indirectly, through state and mutations. The complete history (past, present, future) has no first class representation. Moreover, only discretely evolving values can be (indirectly) captured, since the imperative paradigm is temporally discrete. In contrast, FRP captures these evolving values directly and has no difficulty with continuously evolving values.

FRP is also unusual in that it is concurrent without running afoul of the theoretical & pragmatic rats' nest that plagues imperative concurrency. Semantically, FRP's concurrency is fine-grained, determinate, and continuous. (I'm talking about meaning, not implementation. An implementation may or may not involve concurrency or parallelism.) Semantic determinacy is very important for reasoning, both rigorous and informal. While concurrency adds enormous complexity to imperative programming (due to nondeterministic interleaving), it is effortless in FRP.

那么，什么是FRP? 你可以自己发明的。 从这些想法开始:

Dynamic/evolving values (i.e., values "over time") are first class values in themselves. You can define them and combine them, pass them into & out of functions. I called these things "behaviors". Behaviors are built up out of a few primitives, like constant (static) behaviors and time (like a clock), and then with sequential and parallel combination. n behaviors are combined by applying an n-ary function (on static values), "point-wise", i.e., continuously over time. To account for discrete phenomena, have another type (family) of "events", each of which has a stream (finite or infinite) of occurrences. Each occurrence has an associated time and value. To come up with the compositional vocabulary out of which all behaviors and events can be built, play with some examples. Keep deconstructing into pieces that are more general/simple. So that you know you're on solid ground, give the whole model a compositional foundation, using the technique of denotational semantics, which just means that (a) each type has a corresponding simple & precise mathematical type of "meanings", and (b) each primitive and operator has a simple & precise meaning as a function of the meanings of the constituents. Never, ever mix implementation considerations into your exploration process. If this description is gibberish to you, consult (a) Denotational design with type class morphisms, (b) Push-pull functional reactive programming (ignoring the implementation bits), and (c) the Denotational Semantics Haskell wikibooks page. Beware that denotational semantics has two parts, from its two founders Christopher Strachey and Dana Scott: the easier & more useful Strachey part and the harder and less useful (for software design) Scott part.

如果你坚持这些原则，我希望你能得到或多或少符合FRP精神的东西。

Where did I get these principles? In software design, I always ask the same question: "what does it mean?". Denotational semantics gave me a precise framework for this question, and one that fits my aesthetics (unlike operational or axiomatic semantics, both of which leave me unsatisfied). So I asked myself what is behavior? I soon realized that the temporally discrete nature of imperative computation is an accommodation to a particular style of machine, rather than a natural description of behavior itself. The simplest precise description of behavior I can think of is simply "function of (continuous) time", so that's my model. Delightfully, this model handles continuous, deterministic concurrency with ease and grace.

正确有效地实现这个模型是一个相当大的挑战，但那是另一个故事了。

伙计，这主意太棒了!为什么1998年的时候我没有发现?总之，这是我对Fran教程的理解。建议是最受欢迎的，我正在考虑开始一个基于此游戏引擎。

import pygame
from pygame.surface import Surface
from pygame.sprite import Sprite, Group
from pygame.locals import *
from time import time as epoch_delta
from math import sin, pi
from copy import copy

pygame.init()
screen = pygame.display.set_mode((600,400))
pygame.display.set_caption('Functional Reactive System Demo')

class Time:
    def __float__(self):
        return epoch_delta()
time = Time()

class Function:
    def __init__(self, var, func, phase = 0., scale = 1., offset = 0.):
        self.var = var
        self.func = func
        self.phase = phase
        self.scale = scale
        self.offset = offset
    def copy(self):
        return copy(self)
    def __float__(self):
        return self.func(float(self.var) + float(self.phase)) * float(self.scale) + float(self.offset)
    def __int__(self):
        return int(float(self))
    def __add__(self, n):
        result = self.copy()
        result.offset += n
        return result
    def __mul__(self, n):
        result = self.copy()
        result.scale += n
        return result
    def __inv__(self):
        result = self.copy()
        result.scale *= -1.
        return result
    def __abs__(self):
        return Function(self, abs)

def FuncTime(func, phase = 0., scale = 1., offset = 0.):
    global time
    return Function(time, func, phase, scale, offset)

def SinTime(phase = 0., scale = 1., offset = 0.):
    return FuncTime(sin, phase, scale, offset)
sin_time = SinTime()

def CosTime(phase = 0., scale = 1., offset = 0.):
    phase += pi / 2.
    return SinTime(phase, scale, offset)
cos_time = CosTime()

class Circle:
    def __init__(self, x, y, radius):
        self.x = x
        self.y = y
        self.radius = radius
    @property
    def size(self):
        return [self.radius * 2] * 2
circle = Circle(
        x = cos_time * 200 + 250,
        y = abs(sin_time) * 200 + 50,
        radius = 50)

class CircleView(Sprite):
    def __init__(self, model, color = (255, 0, 0)):
        Sprite.__init__(self)
        self.color = color
        self.model = model
        self.image = Surface([model.radius * 2] * 2).convert_alpha()
        self.rect = self.image.get_rect()
        pygame.draw.ellipse(self.image, self.color, self.rect)
    def update(self):
        self.rect[:] = int(self.model.x), int(self.model.y), self.model.radius * 2, self.model.radius * 2
circle_view = CircleView(circle)

sprites = Group(circle_view)
running = True
while running:
    for event in pygame.event.get():
        if event.type == QUIT:
            running = False
        if event.type == KEYDOWN and event.key == K_ESCAPE:
            running = False
    screen.fill((0, 0, 0))
    sprites.update()
    sprites.draw(screen)
    pygame.display.flip()
pygame.quit()

简而言之:如果每个组成部分都可以被视为一个数字，那么整个系统就可以被视为一个数学方程，对吗?

在阅读了许多页关于FRP的文章后，我终于看到了这篇关于FRP的启发性文章，它最终让我明白了FRP的真正含义。

下面我引用海因里希·阿费尔马斯(活性香蕉的作者)的话。

What is the essence of functional reactive programming? A common answer would be that “FRP is all about describing a system in terms of time-varying functions instead of mutable state”, and that would certainly not be wrong. This is the semantic viewpoint. But in my opinion, the deeper, more satisfying answer is given by the following purely syntactic criterion: The essence of functional reactive programming is to specify the dynamic behavior of a value completely at the time of declaration. For instance, take the example of a counter: you have two buttons labelled “Up” and “Down” which can be used to increment or decrement the counter. Imperatively, you would first specify an initial value and then change it whenever a button is pressed; something like this: counter := 0 -- initial value on buttonUp = (counter := counter + 1) -- change it later on buttonDown = (counter := counter - 1) The point is that at the time of declaration, only the initial value for the counter is specified; the dynamic behavior of counter is implicit in the rest of the program text. In contrast, functional reactive programming specifies the whole dynamic behavior at the time of declaration, like this: counter :: Behavior Int counter = accumulate ($) 0 (fmap (+1) eventUp `union` fmap (subtract 1) eventDown) Whenever you want to understand the dynamics of counter, you only have to look at its definition. Everything that can happen to it will appear on the right-hand side. This is very much in contrast to the imperative approach where subsequent declarations can change the dynamic behavior of previously declared values.

所以，在我的理解中，FRP程序是一组方程:

J是离散的:1,2,3,4…

F依赖于t所以这包含了外部刺激模型的可能性

程序的所有状态都封装在变量x_i中

FRP库考虑了进度时间，换句话说，从j到j+1。

我会在这个视频中更详细地解释这些方程。

编辑:

在最初的回答大约2年后，最近我得出结论，FRP实现还有另一个重要的方面。它们需要(通常也会)解决一个重要的实际问题:缓存失效。

x_i-s的方程描述了一个依赖关系图。当x_i在j时刻发生变化时，并不需要更新j+1时刻的所有其他x_i'值，因此并不需要重新计算所有依赖项，因为有些x_i'可能与x_i无关。

而且，改变的x_i-s可以被增量更新。例如，让我们考虑Scala中的映射操作f=g.map(_+1)，其中f和g是int类型的列表。这里f对应于x_i(t_j) g是x_j(t_j)现在，如果我将一个元素前置到g中，那么对g中的所有元素执行映射操作将是浪费的。一些FRP实现(例如reflect - FRP)旨在解决这个问题。这个问题也称为增量计算。

换句话说，FRP中的行为(x_i-s)可以被认为是缓存的计算。如果某些f_i-s确实发生了变化，FRP引擎的任务就是有效地使这些缓存(x_i-s)失效并重新计算。

Conal Elliott的论文《Simply efficient functional reactivity》(直接PDF, 233 KB)是一个相当好的介绍。相应的库也可以工作。

这篇论文现在被另一篇论文取代，推拉函数式反应性编程(直接PDF, 286 KB)。