In recent implementations of Classic FRP, for instance reactive-banana, there are event streams and signals, which are step functions (reactive-banana calls them behaviours but they are nevertheless step functions). I've noticed that Elm only uses signals, and doesn't differentiate between signals and event streams. Also, reactive-banana allows to go from event streams to signals (edited: and it's sort of possible to act on behaviours using reactimate' although it not considered good practice), which kind of means that in theory we could apply all the event stream combinators on signals/behaviours by first converting the signal to event stream, applying and then converting again. So, given that it's in general easier to use and learn just one abstraction, what is the advantage of having separated signals and event streams ? Is anything lost in using just signals and converting all the event stream combinators to operate on signals ?
edit: The discussion has been very interesting. The main conclusions that I took from the discussion myself is that behaviours/event sources are both needed for mutually recursive definitions (feedback) and for having an output depend on two inputs (a behaviour and an event source) but only cause an action when one of them changes (<#>).
(Clarification: In reactive-banana, it is not possible to convert a Behavior back to an Event. The stepper function is a one-way ticket. There is a changes function, but its type indicates that it is "impure" and it comes with a warning that it does not preserve the semantics.)
I believe that having two separates concepts makes the API more elegant. In other words, it boils down to a question of API usability. I think that the two concepts behave sufficiently differently that things flow better if you have two separate types.
For example, the direct product for each type is different. A pair of Behavior is equivalent to a Behavior of pairs
(Behavior a, Behavior b) ~ Behavior (a,b)
whereas a pair of Events is equivalent to an Event of a direct sum:
(Event a, Event b) ~ Event (EitherOrBoth a b)
If you merge both types into one, then neither of these equivalence will hold anymore.
However, one of the main reasons for the separation of Event and Behavior is that the latter does not have a notion of changes or "updates". This may seem like an omission at first, but it is extremely useful in practice, because it leads to simpler code. For instance, consider a monadic function newInput that creates an input GUI widget that displays the text indicated in the argument Behavior,
input <- newInput (bText :: Behavior String)
The key point now is that the text displayed does not depend on how often the Behavior bText may have been updated (to the same or a different value), only on the actual value itself. This is a lot easier to reason about than the other case, where you would have to think about what happens when two successive event occurrences have the same value. Do you redraw the text while the user edits it?
(Of course, in order to actually draw the text, the library has to interface with the GUI framework and does keep track of changes in the Behavior. This is what the changes combinator is for. However, this can be seen as an optimization and is not available from "within FRP".)
The other main reason for the separation is recursion. Most Events that recursively depend on themselves are ill-defined. However, recursion is always allowed if you have mutual recursion between an Event and a Behavior
e = ((+) <$> b) <#> einput
b = stepper 0 e
There is no need to introduce delays by hand, it just works out of the box.
Something critically important to me is lost, namely the essence of behaviors, which is (possibly continuous) variation over continuous time.
Precise, simple, useful semantics (independent of a particular implementation or execution) is often lost as well.
Check out my answer to "Specification for a Functional Reactive Programming language", and follow the links there.
Whether in time or in space, premature discretization thwarts composability and complicates semantics.
Consider vector graphics (and other spatially continuous models like Pan's). Just as with premature finitization of data structures as explained in Why Functional Programming Matters.
I don't think there's any benefit to using the signals/behaviors abstraction over elm-style signals. As you point out, it's possible to create a signal-only API on top of the signal/behavior API (not at all ready for use, but see https://github.com/JohnLato/impulse/blob/dyn2/src/Reactive/Impulse/Syntax2.hs for an example). I'm pretty sure it's also possible to write a signal/behavior API on top of an elm-style API as well. That would make the two APIs functionally equivalent.
WRT efficiency, with a signals-only API the system should have a mechanism where only signals that have updated values will cause recomputations (e.g. if you don't move the mouse, the FRP network won't re-calculate the pointer coordinates and redraw the screen). Provided this is done, I don't think there's any loss of efficiency compared to a signals-and-streams approach. I'm pretty sure Elm works this way.
I don't think the continuous-behavior issue makes any difference here (or really at all). What people mean by saying behaviors are continuous over time is that they are defined at all times (i.e. they're functions over a continuous domain); the behavior itself isn't a continuous function. But we don't actually have a way to sample a behavior at any time; they can only be sampled at times corresponding to events, so we can't use the full power of this definition!
Semantically, starting from these definitions:
Event == for some t ∈ T: [(t,a)]
Behavior == ∀ t ∈ T: t -> b
since behaviors can only be sampled at times where events are defined, we can create a new domain TX where TX is the set of all times t at which Events are defined. Now we can loosen the Behavior definition to
Behavior == ∀ t ∈ TX: t -> b
without losing any power (i.e. this is equivalent to the original definition within the confines of our frp system). Now we can enumerate all times in TX to transform this to
Behavior == ∀ t ∈ TX: [(t,b)]
which is identical to the original Event definition except for the domain and quantification. Now we can change the domain of Event to TX (by the definition of TX), and the quantification of Behavior (from forall to for some) and we get
Event == for some t ∈ TX: [(t,a)]
Behavior == for some t ∈ TX: [(t,b)]
and now Event and Behavior are semantically identical, so they could obviously be represented using the same structure in an FRP system. We do lose a bit of information at this step; if we don't differentiate between Event and Behavior we don't know that a Behavior is defined at every time t, but in practice I don't think this really matters much. What elm does IIRC is require both Events and Behaviors to have values at all times and just use the previous value for an Event if it hasn't changed (i.e. change the quantification of Event to forall instead of changing the quantification of Behavior). This means you can treat everything as a signal and it all Just Works; it's just implemented so that the signal domain is exactly the subset of time that the system actually uses.
I think this idea was presented in a paper (which I can't find now, anyone else have a link?) about implementing FRP in Java, perhaps from POPL '14? Working from memory, so my outline isn't as rigorous as the original proof.
There's nothing to stop you from creating a more-defined Behavior by e.g. pure someFunction, this just means that within an FRP system you can't make use of that extra defined-ness, so nothing is lost by a more restricted implementation.
As for notional signals such as time, note that it's impossible to implement an actual continuous-time signal using typical programming languages. Since the implementation will necessarily be discrete, converting that to an event stream is trivial.
In short, I don't think anything is lost by using just signals.
I've unfortunately have no references in mind, but I distinctly remember
different reactive authors claiming this choice is just for efficiency. You expose
both to give the programmer a choice in what implementation of the same idea is
more efficient for your problem.
I might be lying now, but I believe Elm implements everything as event streams under the
hood. Things like time wouldn't be so nice like event streams though, since there are an
infinite amount of events during any time frame. I'm not sure how Elm solves this, but I
think it's a good example on something that makes more sense as a signal, both conceptually
and in implementation.
Related
Behaviors are ubiquitously defined as “time-varying value”s1.
Why? time being the dependency/parameter for varying values is very uncommon.
My intuition for FRP would be to have behaviors as event-varying values instead; it is much more common, much more simple, I wage a much more of an efficient idea, and extensible enough to support time too (tick event).
For instance, if you write a counter, you don't care about time/associated timestamps, you just care about the "Increase-button clicked" and "Decrease-button clicked" events.
If you write a game and want a position/force behavior, you just care about the WASD/arrow keys held events, etc. (unless you ban your players for moving to the left in the afternoon; how iniquitous!).
So: Why time is a consideration at all? why timestamps? why are some libraries (e.g. reactive-banana, reactive) take it up to the extent of having Future, Moment values? Why work with event-streams instead of just responding to an event occurrence? All of this just seems to over-complicate a simple idea (event-varying/event-driven value); what's the gain? what problem are we solving here? (I'd love to also get a concrete example along with a wonderful explanation, if possible).
1 Behaviors have been defined so here, here, here... & pretty much everywhere I've encountered.
Behaviors differ from Events primarily in that a Behavior has a value right now while an Event only has a value whenever a new event comes in.
So what do we mean by "right now"? Technically all changes are implemented as push or pull semantics over event streams, so we can only possibly mean "the most recent value as of the last event of consequence for this Behavior". But that's a fairly hairy concept—in practice "now" is much simpler.
The reasoning for why "now" is simpler comes down to the API. Here are two examples from Reactive Banana.
Eventually an FRP system must always produce some kind of externally visible change. In Reactive Banana this is facilitated by the reactimate :: Event (IO ()) -> Moment () function which consumes event streams. There is no way to have a Behavior trigger external changes---you always have to do something like reactimate (someBehavior <# sampleTickEvent) to sample the behavior at concrete times.
Behaviors are Applicatives unlike Events. Why? Well, let's assume Event was an applicative and think about what happens when we have two event streams f and x and write f <*> x: since events occur all at different times the chances of f and x being defined simultaneously are (almost certainly) 0. So f <*> x would always mean the empty event stream which is useless.
What you really want is for f <*> x to cache the most current values for each and take their combined value "all of the time". That's really confusing concept to talk about in terms of an event stream, so instead lets consider f and x as taking values for all points in time. Now f <*> x is also defined as taking values for all points in time. We've just invented Behaviors.
Because it was the simplest way I could think of to give a precise denotation (implementation-independent meaning) to the notion of behaviors, including the sorts of operations I wanted, including differentiation and integration, as well as tracking one or more other behaviors (including but not limited to user-generated behavior).
Why? time being the dependency/parameter for varying values is very uncommon.
I suspect that you're confusing the construction (recipe) of a behavior with its meaning. For instance, a behavior might be constructed via a dependency on something like user input, possibly with additional synthetic transformation. So there's the recipe. The meaning, however, is simply a function of time, related to the time-function that is the user input. Note that by "function", I mean in the math sense of the word: a (deterministic) mapping from domain (time) to range (value), not in the sense that there's a purely programmatic description.
I've seen many questions asking why time matters and why continuous time. If you apply the simple discipline of giving a mathematical meaning in the style of denotational semantics (a simple and familiar style for functional programmers), the issues become much clearer.
If you really want to grok the essence of and thinking behind FRP, I recommend you read my answer to "Specification for a Functional Reactive Programming language" and follow pointers, including "What is Functional Reactive Programming".
Conal Elliott's Push-Pull FRP paper describes event-varying data, where the only points in time that are interesting are when events occcur. Reactive event-varying data is the current value and the next Event that will change it. An Event is a Future point in the event-varying Reactive data.
data Reactive a = a ‘Stepper ‘ Event a
newtype Event a = Ev (Future (Reactive a))
The Future doesn't need to have a time associated with it, it just need to represent the idea of a value that hasn't happened yet. In an impure language with events, for example, a future can be an event handle and a value. When the event occurs, you set the value and raise the handle.
Reactive a has a value for a at all points in time, so why would we need Behaviors? Let's make a simple game. In between when the user presses the WASD keys, the character, accelerated by the force applied, still moves on the screen. The character's position at different points in time is different, even though no event has occurred in the intervening time. This is what a Behavior describes - something that not only has a value at all points in time, but its value can be different at all points in time, even with no intervening events.
One way to describe Behaviors would be to repeat what we just stated. Behaviors are things that can change in-between events. In-between events they are time-varying values, or functions of time.
type Behavior a = Reactive (Time -> a)
We don't need Behavior, we could simply add events for clock ticks, and write all of the logic in our entire game in terms of these tick events. This is undesirable to some developers as the code declaring what our game is is now intermingled with the code providing how it is implemented. Behaviors allow the developer to separate this logic between the description of the game in terms of time-varying variables and the implementation of the engine that executes that description.
I have scoured the documentation of reactive-banana, and I can't find a way to specify explicit time delays. Say, for example, I would like to take an Event t a and shift all of its occurrences 1 second into the future; or get an event that fires 1 second from now (within Moment t); or anything like that. Are explicit delays representable in reactive-banana? If not, how do users implement, e.g., echoing input delayed by a second?
As Ben indicates, this is correct: reactive-banana is no built-in notion of time and delays.
The main reason is that it is hard to guarantee that logical time and real time agree. What happens when a mouse click happens in real time before the logical time of an event which could not yet be calculated, i.e. whose real time dallies behind its logical time?
However, it is still possible to deal with time by using external timer events. Have a look at the Wave.hs example.
I'm not an expert by any means, but I've used an older reactive-banana for a program with exactly this need (I don't have the code on hand at the moment, unfortunately).
I think that explicit delays don't fit super-naturally within the reactive-banana framework because you're always implicitly dealing with some sort of abstract time. Delaying by particular intervals requires dealing with the real world notion of time. I can see that it would be very hard to say what something like union event (delay 1.0 event) would mean in terms of the authoritative model where type Event a = [Maybe a]: should it be [a, b, a, b] or [a, a, b, b]? As I've been thinking of it, the problem is that which of those two should result is unspecified without considering operational runtime details.
But you can do it "around" the reactive-banana framework by wiring up an IO action to the event you want to delay that sleeps for a while (not in the main thread, obviously) and then fires a different event. I believe I was able to wrap all that logic up in a function I could call from the NetworkDescription monad (this was before the Moment monad existed) to get my new delayed event.
Pardon me, I'm just starting to look into reactive-banana and FRP.
The author of reactive-banana made this example per my suggestion, in which he creates a counter which can be increased and decreased. He uses accumE function which accumulates events. I think I was able to somewhat grok the Event type, and was able to test quite a few things with it, but then I remembered that there was also Behavior. I looked into it, but it seems like the behavior is meant to be used in similar situations; to modify an existing variable, just like accumE does with events.
What does Behavior mean, and what are the use cases for it?
I agree with Ankur rather than Chris: a text box is a value over time and so naturally wants to be a behavior rather than an event. The reasons Chris give for the less natural choice of event are implementation issues and so (if accurate) an unfortunate artifact of the reactive-banana implementation. I'd much rather see the implementation improved than the paradigm used unnaturally.
Besides the semantic fit, it's pragmatically very useful to choose Behavior over Event. You can then, for instance, use the Applicative operations (e.g., liftA2) to combine the time-varying text box value with other time-varying values (behaviors).
Semantically, you have
Behavior a = Time -> a
That is, a Behavior a is a value of type a that varies over time. In general, you know nothing at all about when a Behavior a would change, so it turns out to be a rather poor choice for updating a text field on the click of a button. That said, it would be easy to get a behavior that expresses the current value of the number in the counter example. Just use stepper on the event stream, or alternatively, build it from scratch the same way, except by using accumB instead of accumE.
Typically, things you hook up to input and output will always be Events, so Behavior is used internally for intermediate results.
Suppose that in the given example, you want to add a new button that remembers the current value, like the memory function on simple calculators. You would start out by adding a memory button and a text field for the remembered value:
bmem <- button f [text := "Remember"]
memory <- staticText f []
You need to be able to ask for the current value at any time, so in your network, you'd add a behavior to represent it.
let currentVal = stepper 0 counter
Then you can hook up events, and use apply to read the value of the behavior every time the Remember button is pressed, and produce an Event with that sequence of values.
emem <- event0 bmem command
let memoryE = apply (const <$> currentVal) emem
And finally, hook up this new event to the output
sink memory [text :== ("", show <$> memoryE)]
If you wanted to use memory internally, then again you'd want a Behavior for its current value too... but since we only ever use it to connect it to an output, we only need an event for now.
Does that help?
Library author speaking. :-)
Apparently, Chris Smith can read minds because he accurately describes what I am thinking. :-)
But Conal and Arthur have a point, too. Conceptually, the counter is a value that varies in time, not a sequence of event occurrences. Thus, thinking of it as a Behavior would be more appropriate.
Unfortunately, behaviors do not come with any information about when they will change, the are "poll-only". Now, I could try to implement various clever schemes that will minimize the polling and thus allow effient updates of GUI elements. (Conal does something similar in the original paper.) But I have adopted a "no magic" philosophy: the library user shall be responsible for managing updates via events himself.
The solution I currently envision is to provide a third type besides Event and Behavior, namely Reactive (name subject to change) which embodies qualities of both: conceptually, it's a value that varies in time, but it also comes with an event that notifies of changes. One possible implementation would be
type Reactive a = (a,Event a)
changes :: Reactive a -> Event a
changes (_, e) = e
value :: Reactive a -> Behavior a
value (x, e) = stepper x e
It is no surprise that this is precisely the type that sink expects. This will be included in a future version of the reactive-banana library.
EDIT: I have released reactive-banana version 0.4 which includes the new type, which is now called Discrete.
Generally, Behavior is a value that changes over a period of time. It is a continuous value, where as events are discrete values. In case of Behavior a value is always present.
For example: The text on a text box is a Behavior as the text can change over a period of time but there will be a current value, where as a keyboard stroke in a event as you cannot query a keyboard stroke for its "current" value.
I am trying to figure out how to do the following, assume that your are working on a controller for a DC motor you want to keep it spinning at a certain speed set by the user,
(def set-point (ref {:sp 90}))
(while true
(let [curr (read-speed)]
(controller #set-point curr)))
Now that set-point can change any time via a web a application, I can't think of a way to do this without using ref, so my question is how functional languages deal with this sort of thing? (even though the example is in clojure I am interested in the general idea.)
This will not answer your question but I want to show how these things are done in Clojure. It might help someone reading this later so they don't think they have to read up on monads, reactive programming or other "complicated" subjects to use Clojure.
Clojure is not a purely functional language and in this case it might be a good idea to leave the pure functions aside for a moment and model the inherent state of the system with identities.
In Clojure, you would probably use one of the reference types. There are several to choose from and knowing which one to use might be difficult. The good news is they all support the unified update model so changing the reference type later should be pretty straight forward.
I've chosen an atom but depending on your requirements it might be more appropriate to use a ref or an agent.
The motor is an identity in your program. It is a "label" for some thing that has different values at different times and these values are related to each other (i.e., the speed of the motor). I have put a :validator on the atom to ensure that the speed never drops below zero.
(def motor (atom {:speed 0} :validator (comp not neg? :speed)))
(defn add-speed [n]
(swap! motor update-in [:speed] + n))
(defn set-speed [n]
(swap! motor update-in [:speed] (constantly n)))
> (add-speed 10)
> (add-speed -8)
> (add-speed -4) ;; This will not change the state of motor
;; since the speed would drop below zero and
;; the validator does not allow that!
> (:speed #motor)
2
> (set-speed 12)
> (:speed #motor)
12
If you want to change the semantics of the motor identity you have at least two other reference types to choose from.
If you want to change the speed of the motor asynchronously you would use an agent. Then you need to change swap! with send. This would be useful if, for example, the clients adjusting the motor speed are different from the clients using the motor speed, so that it's fine for the speed to be changed "eventually".
Another option is to use a ref which would be appropriate if the motor need to coordinate with other identities in your system. If you choose this reference type you change swap! with alter. In addition, all state changes are run in a transaction with dosync to ensure that all identities in the transaction are updated atomically.
Monads are not needed to model identities and state in Clojure!
For this answer, I'm going to interpret "a purely functional language" as meaning "an ML-style language that excludes side effects" which I will interpret in turn as meaning "Haskell" which I'll interpret as meaning "GHC". None of these are strictly true, but given that you're contrasting this with a Lisp derivative and that GHC is rather prominent, I'm guessing this will still get at the heart of your question.
As always, the answer in Haskell is a bit of sleight-of-hand where access to mutable data (or anything with side effects) is structured in such a way that the type system guarantees that it will "look" pure from the inside, while producing a final program that has side effects where expected. The usual business with monads is a large part of this, but the details don't really matter and mostly distract from the issue. In practice, it just means you have to be explicit about where side effects can occur and in what order, and you're not allowed to "cheat".
Mutability primitives are generally provided by the language runtime, and accessed through functions that produce values in some monad also provided by the runtime (often IO, sometimes more specialized ones). First, let's take a look at the Clojure example you provided: it uses ref, which is described in the documentation here:
While Vars ensure safe use of mutable storage locations via thread isolation, transactional references (Refs) ensure safe shared use of mutable storage locations via a software transactional memory (STM) system. Refs are bound to a single storage location for their lifetime, and only allow mutation of that location to occur within a transaction.
Amusingly, that whole paragraph translates pretty directly to GHC Haskell. I'm guessing that "Vars" are equivalent to Haskell's MVar, while "Refs" are almost certainly equivalent to TVar as found in the stm package.
So to translate the example to Haskell, we'll need a function that creates the TVar:
setPoint :: STM (TVar Int)
setPoint = newTVar 90
...and we can use it in code like this:
updateLoop :: IO ()
updateLoop = do tvSetPoint <- atomically setPoint
sequence_ . repeat $ update tvSetPoint
where update tv = do curSpeed <- readSpeed
curSet <- atomically $ readTVar tv
controller curSet curSpeed
In actual use my code would be far more terse than that, but I've left things more verbose here in hopes of being less cryptic.
I suppose one could object that this code isn't pure and is using mutable state, but... so what? At some point a program is going to run and we'd like it to do input and output. The important thing is that we retain all the benefits of code being pure, even when using it to write code with mutable state. For instance, I've implemented an infinite loop of side effects using the repeat function; but repeat is still pure and behaves reliably and nothing I can do with it will change that.
A technique to tackle problems that apparently scream for mutability (like GUI or web applications) in a functional way is Functional Reactive Programming.
The pattern you need for this is called Monads. If you really want to get into functional programming you should try to understand what monads are used for and what they can do. As a starting point I would suggest this link.
As a short informal explanation for monads:
Monads can be seen as data + context that is passed around in your program. This is the "space suit" often used in explanations. You pass data and context around together and insert any operation into this Monad. There is usually no way to get the data back once it is inserted into the context, you just can go the other way round inserting operations, so that they handle data combined with context. This way it almost seems as if you get the data out, but if you look closely you never do.
Depending on your application the context can be almost anything. A datastructure that combines multiple entities, exceptions, optionals, or the real world (i/o-monads). In the paper linked above the context will be execution states of an algorithm, so this is quite similar to the things you have in mind.
In Erlang you could use a process to hold the value. Something like this:
holdVar(SomeVar) ->
receive %% wait for message
{From, get} -> %% if you receive a get
From ! {value, SomeVar}, %% respond with SomeVar
holdVar(SomeVar); %% recursively call holdVar
%% to start listening again
{From, {set, SomeNewVar}} -> %% if you receive a set
From ! {ok}, %% respond with ok
holdVar(SomeNewVar); %% recursively call holdVar with
%% the SomeNewVar that you received
%% in the message
end.
Many a times, I've come across statements of the form
X does/doesn't compose well.
I can remember few instances that I've read recently :
Macros don't compose well (context: clojure)
Locks don't compose well (context: clojure)
Imperative programming doesn't compose well... etc.
I want to understand the implications of composability in terms of designing/reading/writing code ? Examples would be nice.
"Composing" functions basically just means sticking two or more functions together to make a big function that combines their functionality in a useful way. Essentially, you define a sequence of functions and pipe the results of each one into the next, finally giving the result of the whole process. Clojure provides the comp function to do this for you, you could do it by hand too.
Functions that you can chain with other functions in creative ways are more useful in general than functions that you can only call in certain conditions. For example, if we didn't have the last function and only had the traditional Lisp list functions, we could easily define last as (def last (comp first reverse)). Look at that — we didn't even need to defn or mention any arguments, because we're just piping the result of one function into another. This would not work if, for example, reverse took the imperative route of modifying the sequence in-place. Macros are problematic as well because you can't pass them to functions like comp or apply.
Composition in programming means assembling bigger pieces out of smaller ones.
Composition of unary functions creates a more complicated unary function by chaining simpler ones.
Composition of control flow constructs places control flow constructs inside other control flow constructs.
Composition of data structures combines multiple simpler data structures into a more complicated one.
Ideally, a composed unit works like a basic unit and you as a programmer do not need to be aware of the difference. If things fall short of the ideal, if something doesn't compose well, your composed program may not have the (intended) combined behavior of its individual pieces.
Suppose I have some simple C code.
void run_with_resource(void) {
Resource *r = create_resource();
do_some_work(r);
destroy_resource(r);
}
C facilitates compositional reasoning about control flow at the level of functions. I don't have to care about what actually happens inside do_some_work(); I know just by looking at this small function that every time a resource is created on line 2 with create_resource(), it will eventually be destroyed on line 4 by destroy_resource().
Well, not quite. What if create_resource() acquires a lock and destroy_resource() frees it? Then I have to worry about whether do_some_work acquires the same lock, which would prevent the function from finishing. What if do_some_work() calls longjmp(), and skips the end of my function entirely? Until I know what goes on in do_some_work(), I won't be able to predict the control flow of my function. We no longer have compositionality: we can no longer decompose the program into parts, reason about the parts independently, and carry our conclusions back to the whole program. This makes designing and debugging much harder and it's why people care whether something composes well.
"Bang for the Buck" - composing well implies a high ratio of expressiveness per rule-of-composition. Each macro introduces its own rules of composition. Each custom data structure does the same. Functions, especially those using general data structures have far fewer rules.
Assignment and other side effects, especially wrt concurrency have even more rules.
Think about when you write functions or methods. You create a group of functionality to do a specific task. When working in an Object Oriented language you cluster your behavior around the actions you think a distinct entity in the system will perform. Functional programs break away from this by encouraging authors to group functionality according to an abstraction. For example, the Clojure Ring library comprises a group of abstractions that cover routing in web applications.
Ring is composable where functions that describe paths in the system (routes) can be grouped into higher order functions (middlewhere). In fact, Clojure is so dynamic that it is possible (and you are encouraged) to come up with patterns of routes that can be dynamically created at runtime. This is the essence of composablilty, instead of coming up with patterns that solve a certain problem you focus on patterns that generate solutions to a certain class of problem. Builders and code generators are just two of the common patterns used in functional programming. Function programming is the art of patterns that generate other patterns (and so on and so on).
The idea is to solve a problem at its most basic level then come up with patterns or groups of the lowest level functions that solve the problem. Once you start to see patterns in the lowest level you've discovered composition. As folks discover second order patterns in groups of functions they may start to see a third level. And so on...
Composition (in the context you describe at a functional level) is typically the ability to feed one function into another cleanly and without intermediate processing. Such an example of composition is in std::cout in C++:
cout << each << item << links << on;
That is a simple example of composition which doesn't really "look" like composition.
Another example with a form more visibly compositional:
foo(bar(baz()));
Wikipedia Link
Composition is useful for readability and compactness, however chaining large collections of interlocking functions which can potentially return error codes or junk data can be hazardous (this is why it is best to minimize error code or null return values.)
Provided your functions use exceptions, or alternatively return null objects you can minimize the requirement for branching (if) on errors and maximize the compositional potential of your code at no extra risk.
Object composition (vs inheritance) is a separate issue (and not what you are asking, but it shares the name). It is one of containment to derive object hierarchy as opposed to direct inheritance.
Within the context of clojure, this comment addresses certain aspects of composability. In general, it seems to emerge when units of the system do one thing well, do not require other units to understand its internals, eschew side-effects, and accept and return the system's pervasive data structures. All of the above can be seen in M2tM's C++ example.
composability, applied to functions, means that the functions are smaller and well-defined, thus easy to integrate into other functions (i have seen this idea in the book "the joy of clojure")
the concept can apply to other things that are supposed be composed into something else.
the purpose of composability is reuse. for example, a function well-build (composable) is easier to reuse
macros aren't that well-composable because you can't pass them as parameters
lock are crap because you can't really give them names (define them well) or reuse them. you just do them inplace
imperative languages aren't that composable because (some of them, at least) don't have closures. if you want functionality passed as parameter, you're screwed. you have to build an object and pass that; disclaimer here: this last idea i'm not entirely convinced is true, therefore research more before taking it for granted
another idea on imperative languages is that they don't compose well because they imply state (from wikipedia knowledgebase :) "Imperative programming - describes computation in terms of statements that change a program state").
state does not compose well because although you have given a specific "something" in input, that "something" generates an output according to it's state. different internal state, different behaviour. and thus you can say good-bye to what you where expecting to happen.
with state, you depend to much on knowing what the current state of an object is... if you want to predict it's behavior. more stuff to keep in the back of your mind, less composable (remember well-defined ? or "small and simple", as in "easy to use" ?)
ps: thinking of learning clojure, huh ? investigating... ? good for you ! :P