Function.schelp

class::Function
summary::Implements a function
categories::Core>Kernel
related::Classes/FunctionDef

description::
A Function is a reference to a FunctionDef and its defining context Frame. When a FunctionDef is encountered in your code it is pushed on the stack as a Function. A Function can be evaluated by using the 'value' method. See the Functions help file for a basic introduction.

Because it inherits from AbstractFunction, Functions respond to math operations by creating a new Function.

code::
// example
(
var a, b, c;
a = { [100, 200, 300].choose };	// a Function
b = { 10.rand + 1 };	// another Function
c = a + b; 	// c is a Function.
c.value.postln;	// evaluate c and print the result
)
::

See link::Classes/AbstractFunction#Function Composition:: for function composition examples.

Because Functions are such an important concept, here some examples from related programming languages with functions as first class objects:

code::
// returning the first argument itself:
{ |x| x }.value(1) // SuperCollider
[:x | x ] value: 1 // Smalltalk
((lambda (x) x) 1) // Lisp
::

subsection:: Related Keywords

method:: thisFunction
The global pseudo-variable code::thisFunction:: always evaluates to the current
enclosing Function.

note::
Be aware of link::Reference/Control-Structures#Inline
optimization#inline optimizations:: which will be reflected in the value of
code::thisFunction::.
::

See also: link::Classes/FunctionDef#.thisFunctionDef#thisFunctionDef::


classMethods::

private::new

instancemethods::

subsection::Access

method::def

Get the definition ( FunctionDef ) of the Function.

method::isClosed

returns true if the function is closed, i.e. has no external references and can thus be converted to a compile string safely.

subsection::Evaluation

method::value

Evaluates the FunctionDef referred to by the Function. The Function is passed the args given.

code::
{ |a, b| (a * b).postln }.value(3, 10);
{ arg a, b; (a * b).postln }.value(3, 10); // different way of expressing the same
::

method::valueArray

Evaluates the FunctionDef referred to by the Function. If the last argument is an Array or List, then it is unpacked and appended to the other arguments (if any) to the Function. If the last argument is not an Array or List then this is the same as the 'value' method.

code::
{ |a, b, c| ((a * b) + c).postln }.valueArray([3, 10, 7]);

{ |a, b, c, d| [a, b, c, d].postln }.valueArray([1, 2, 3]);

{ |a, b, c, d| [a, b, c, d].postln }.valueArray(9, [1, 2, 3]);

{ |a, b, c, d| [a, b, c, d].postln }.valueArray(9, 10, [1, 2, 3]);
::

A common syntactic shortcut:

code::
{ |a, b, c| ((a * b) + c).postln }.value(*[3, 10, 7]);
::

method::valueEnvir

As value above. Unsupplied argument names are looked up in the current Environment.

code::
(
Environment.use({
~a = 3;
~b = 10;
{ |a, b| (a * b).postln }.valueEnvir;
});
)
::

method::valueArrayEnvir

Evaluates the FunctionDef referred to by the Function. If the last argument is an Array or List, then it is unpacked and appended to the other arguments (if any) to the Function. If the last argument is not an Array or List then this is the same as the 'value' method. Unsupplied argument names are looked up in the current Environment.


method::valueWithEnvir

Evaluate the function, using arguments from the supplied environment
This is slightly faster than valueEnvir and does not require replacing the currentEnvironment

code::
(
e = Environment.make({ ~a = 3; ~b = 10 });
{ |a, b| (a * b) }.valueWithEnvir(e);
)
::

method::functionPerformList

For Function, this behaves the same as valueArray(arglist). It is used  where Functions and other objects should behave differently to value, such as in the object prototyping implementation of Environment.


method::performWithEnvir

code::
a = { |a, b, c| postf("% plus % plus % is %\n", a, b, c, a + b + c); "" };
a.performWithEnvir(\value, (a: 1, c: 3, d: 4, b: 2));
::

argument::selector
A Symbol representing a method selector.
argument::envir
The remaining arguments derived from the environment and passed as arguments to the method named by the selector.

method::performKeyValuePairs

code::
a = { |a, b, c| postf("% plus % plus % is %\n", a, b, c, a + b + c); "" };
a.performKeyValuePairs(\value, [\a, 1, \b, 2, \c, 3, \d, 4]);
::

argument::selector
A Symbol representing a method selector.
argument::pairs
Array or List with key-value pairs.


method::loop

Repeat this function. Useful with Task and Clocks.

code::
t = Task({ { "I'm loopy".postln; 1.wait;}.loop });
t.start;
t.stop;
::

method::defer

Delay the evaluation of this Function by code::delta:: in seconds on AppClock.

This is equivalent to code::AppClock.sched(0, function):: unless code::delta:: is code::nil::. In that case the function is only scheduled if current code is not running on AppClock, otherwise the function is evaluated immediately.

code::
{ "2 seconds have passed.".postln; }.defer(2);

(
{ "chicken".postln }.defer(0); // schedules on the AppClock
{ "egg".postln }.defer // evaluates immediately
)

(
fork { // schedules on a TempoClock
    { "chicken".postln }.defer // schedules on the AppClock
};
{ "egg".postln }.defer // evaluates immediately
)
::

method::dup

Return an Array consisting of the results of n evaluations of this Function.

code::
x = { 4.rand; }.dup(4);
x.postln;
::

method::!

equivalent to dup(n)

code::
x = { 4.rand } ! 4;
x.postln;
::

method::sum

return the sum of n values produced.

code::
{ 4.rand }.sum(8);
::

method::choose

evaluates the function. This makes it polymorphic to SequenceableCollection, Bag and Set.

code::
[{ 100.rand }, [20, 30, 40]].collect(_.choose);
::

method::bench

Returns the amount of time this function takes to evaluate. print is a boolean indicating whether the result is posted. The default is true.

code::
{ 1000000.do({ 1.0.rand }); }.bench;
::

method::fork

Returns a Routine using the receiver as it's function, and plays it in a TempoClock.

code::
{ 4.do({ "Threading...".postln; 1.wait;}) }.fork;
::

method::forkIfNeeded

If needed, creates a new Routine to evaluate the function in, if the message is called within a routine already, it simply evaluates it.

code::
f = { 4.do({ "Threading...".postln; 1.wait;}) };
f.forkIfNeeded;
{ "we are now in a routine".postln; 1.wait; f.forkIfNeeded }.fork;
::

method::block

Break from a loop. Calls the receiver with an argument which is a function that returns from the method block. To exit the loop, call .value on the function passed in. You can pass a value to this function and that value will be returned from the block method.

code::
block {|break|
	100.do {|i|
		i.postln;
		if (i == 7) { break.value(999) }
	};
}
::

method::thunk

Return a Thunk, which is an unevaluated value that can be used in calculations

code::
x = thunk { 4.rand };
x.value;
x.value;
::

method::flop

Return a function that, when evaluated with nested arguments, does multichannel expansion by evaluating the receiver function for each channel. A flopped function responds like the "map" function in languages like Lisp.

code::
f = { |a, b| if(a > 0) { a + b } { -inf } }.flop;
f.value([-1, 2, 1, -3.0], [10, 1000]);
f.value(2, 3);
::


method::envirFlop

like flop, but implements an environment argument passing (valueEnvir).
Less efficient in generation than flop, but not a big difference in evaluation.

code::
f = { |a| if(a > 0) { a + 1 } { -inf } }.envirFlop;
e = (a: [20, 40]);
e.use { f.value }
::


method::inEnvir

returns an "environment-safe" function. See Environment for more details.

code::
// prints nil because ~a is read from topEnvironment, not e
e = (a: "got it", f: { { ~a.postln }.defer(0.5) });
e.use { e.f };

// prints "got it" because { ~a.postln } is now bound to the e environment
e = (a: "got it", f: { { ~a.postln }.inEnvir.defer(0.5) });
e.use { e.f };
::


method::case

Function implements a case method which allows for conditional evaluation with multiple cases. Since the receiver represents the first case this can be simply written as pairs of test functions and corresponding functions to be evaluated if true. Unlike Object-switch, this is inlined and is therefore just as efficient as nested if statements.

code::
(
var i, x, z;
z = [0, 1, 1.1, 1.3, 1.5, 2];
i = z.choose;
x = case
	{ i == 1 }   { \no }
	{ i == 1.1 } { \wrong }
	{ i == 1.3 } { \wrong }
	{ i == 1.5 } { \wrong }
	{ i == 2 }   { \wrong }
	{ i == 0 }   { \true };
x.postln;
)
::

method::matchItem

Interface shared with other classes that implements pattern matching. See also: matchItem.
Function.matchItem evaluates the function with the item as argument, expecting a Boolean as reply.

See also link::Reference/matchItem::.

code::
{ |x| x > 5 }.matchItem(6); // true
::


method::performDegreeToKey

use a function as a conversion from scale degree to note number. See also SequenceableCollection and Scale

code::
// a strange mapping
(
var f = {|degree, stepsPerOctave, acc|
	(1.8 ** (degree % stepsPerOctave) + acc).postln
};
Pbind(
	\scale, f,
	\degree, Pseq([0, 1, 2b, 3s, 4s, 6, 14, [0, 2, 4], [1, 3, 6]], inf)
).play
)
::

subsection::Exception Handling


For the following two methods a return ^ inside of the receiver itself cannot be caught. Returns in methods called by the receiver are OK.


method::try

Executes the receiver. If an exception is thrown the catch function handler is executed with the error as an argument. handler itself can rethrow the error if desired.

method::protect

Executes the receiver. The cleanup function handler is executed with an error as an argument, or nil if there was no error. The error continues to be in effect.

subsection:: Scheduling


method:: awake

This method is called by a link::Classes/Clock:: on which the function was
scheduled when its scheduling time is up. It calls link::#-value::, passing
on the scheduling time in beats as an argument. 

argument:: beats
The scheduling time in beats. This is equal to the current logical time
(link::Classes/Thread#-beats::).

argument:: seconds
The scheduling time in seconds. This is equal to the current logical time
(link::Classes/Thread#-seconds::).

argument:: clock
The clock on which the object was scheduled.

returns::
The value returned by the function's link::#-value::. A caller clock uses
this value to reschedule the function.

code::
// Runs every 2 seconds
AppClock.play({ "And again".postln; 2 });
::

subsection::Audio

method::play

Play a Synth from UGens returned by the function. The function arguments become controls that can be set afterwards.

This works as follows: play wraps the UGens in a SynthDef and sends it to the target. When this asynchronous command is completed, it creates one synth from this definition.


argument::target
A Node, Server, or Nil. A Server will be converted to the default group of that server. Nil will be converted to the default group of the default Server.

argument::outbus
A bus number that determines the output channel. This is equivalent to code::Out.ar(outbus, signal)::. The default is 0.

argument::fadeTime
Crossfade time for attack and release of the synth. The default is 0.02 seconds, which is just enough to avoid a click.

argument::addAction
The add action for the synth.  For a list of valid addActions see link::Classes/Synth::

argument::args
An array of key value pairs of control names and control values which are used when starting the synth.

note::
Some UGens are added in this process.
list::
## an link::Classes/Out:: UGen for playing the audio to the first audio busses. If the function returns an Out UGen, this is omitted. ##an envelope with a code::\gate:: control for releasing and crossfading. If the function provides its own releasable envelope, this is omitted.
::
::




code::
x = { |freq = 440| SinOsc.ar(freq, 0, 0.3) }.play; // this returns a Synth object;
x.set(\freq, 880); // note you can set the freq argument
x.defName; // the name of the resulting SynthDef
x.release(4); // fadeout over 4 seconds
::

Many of the examples make use of the Function.play syntax.
Note that reusing such code in a SynthDef requires the addition of an Out ugen.

code::
// the following two lines produce equivalent results
{ SinOsc.ar(440, 0, 0.3) }.play(fadeTime: 0.0);
SynthDef(\help_FuncPlay, { Out.ar(0, SinOsc.ar(440, 0, 0.3))}).play;
::

Function.play is often more convenient than SynthDef.play, particularly for short examples and quick testing. The latter does have some additional options, such as lagtimes for controls, etc. Where reuse and maximum flexibility are of greater importance, SynthDef and its various methods are usually the better choice.


method::scope

As play above, and calls Server-scope to open a scope window in which to view the output.

code::
{ FSinOsc.ar(440, 0, 0.3) }.scope(1)
::

argument::numChannels
The number of channels to display in the scope window, starting from outbus. The default is 2.
argument::outbus
The output bus to play the audio out on. This is equivalent to Out.ar(outbus, theoutput). The default is 0.
argument::fadeTime
A fadein time. The default is 0.02 seconds, which is just enough to avoid a click.
argument::bufsize
The size of the buffer for the ScopeView. The default is 4096.
argument::zoom
A zoom value for the scope's X axis. Larger values show more. The default is 1.

method::plot

note::See link::Reference/plot:: for a general reference on plotting::


Calculates duration in seconds worth of the output of this function asynchronously, and plots it in a GUI window. Unlike play and scope it will not work with explicit Out Ugens, so your function should return a UGen or an Array of them. The plot will be calculated in realtime.

code::
{ SinOsc.ar(440) }.plot(0.01, bounds: Window.screenBounds);

{ {|i| SinOsc.ar(1 + i)}.dup(7) }.plot(1);
::



argument::duration
The duration of the function to plot in seconds. The default is 0.01.
argument::target
The server, synth or group where the signal is picked up. See link::Reference/asTarget::.
code::
(
var synth = { Out.ar(99, SinOsc.ar([300, 700])) }.play;
{ In.ar(99, 2) }.plot(0.02, synth);
)
::

argument::bounds
An instance of Rect or Point indicating the bounds of the plot window.
argument::minval
the minimum value in the plot. If not specified, the signal's minimum is used.
argument::maxval
the maximum value in the plot. If not specified, the signal's maximum is used.
argument:: separately
If set to code::true::, for multichannel signals use separate value display ranges, unless minval or maxval is specified already.


method::asBuffer
Calculates duration in seconds worth of the output of this function asynchronously, and returns it in a link::Classes/Buffer:: of the number of channels. This method immediately returns a buffer, which takes code::duration:: seconds to become filled with data asynchronously. Then the action function is called.

argument::duration
The duration of the function to plot in seconds. The default is 0.01.

argument::target
The synth or group after which the synth runs (see link::Reference/asTarget::). The default is the default group of code::Server.default::.

argument::action
A function that is called when the buffer is filled. It is passed the buffer as argument.

argument::fadeTime
A fade in and out time in seconds. Only when greater than zero, an envelope is applied to the signal to avoid clicks (default: 0).

discussion::
code::
// record a buffer
b = { Blip.ar(XLine.kr(10000, 4, 3) * [1, 1.2], 20) * 0.1 }.asBuffer(3, fadeTime:0.1)
b.plot; // after 3 seconds, you can see it.
// play the soundfile back
{ PlayBuf.ar(b.numChannels, b, LFNoise2.kr(2 ! 8).exprange(0.15, 1), loop:1).sum }.play;
::


method:: loadToFloatArray
Write a UGen function to a file and then load it into a FloatArray.

argument:: duration
Amount of data to record in seconds

argument::target
A server, synth, or group where to place the synth that plays the function

argument::action
A function to call when writing is finished.

code::
// get some data from the synth audio
{ SinOsc.ar + GrayNoise.ar(0.1) }.loadToFloatArray(0.01, action: { arg array; a = array });
a.postln;
a.plot;

// place it after some other node:
z = { WhiteNoise.ar * 0.1 }.play; // play some noise
{ BPF.ar(In.ar(0), 7000, 0.01) * 10 }.loadToFloatArray(0.01, z, action: { arg array; a = array });
a.postln;
::

method:: getToFloatArray
Stream the function audio values to the client using a series of getn messages and put the results into a FloatArray.

argument:: duration
Amount of data to record in seconds

argument::target
A server, synth, or group where to place the synth that plays the function

argument::action
A function to call when writing is finished.

argument:: wait
The amount of time in seconds to wait between sending getn messages. Longer times are safer. The default is 0.01 seconds which seems reliable under normal circumstances. A setting of 0 is not recommended.

argument:: timeout
The amount of time in seconds after which to post a warning if all replies have not yet been received. the default is 3.

discussion::
For a discussion of the difference between getToFloatArray and loadToFloatArray, see link::Classes/Buffer#-getToFloatArray:: and link::Classes/Buffer#-loadToFloatArray::.


code::
// get some data from the synth audio

{ SinOsc.ar + GrayNoise.ar(0.1) }.getToFloatArray(0.01, action: { arg array; a = array });
a.postln;
a.plot;
::


subsection::Conversion

method::asSynthDef

Returns a SynthDef based on this Function, adding a Linen and an Out ugen if needed.

argument::rates
An Array of rates and lagtimes for the function's arguments (see SynthDef for more details).
argument::prependArgs
arguments
argument::outClass
The class of the output ugen as a symbol. The default is \Out.
argument::fadeTime
a fadein time. The default is 0.
argument::name
the name of the SynthDef

method::asDefName

Performs asSynthDef (see above), sends the resulting def to the local server and returns the SynthDefs name. This is asynchronous.

code::
x = { SinOsc.ar(440, 0, 0.3) }.asDefName; // this must complete first
y = Synth(x);
::

method::asRoutine

Returns a Routine using this as its func argument.

method::r

Returns a Routine using this as its func argument.

code::
a = r { 5.do { |i| i.rand.yield } };
a.nextN(8);
::

method::p

Returns a Prout using this as its func argument.

code::
a = p { 5.do { |i| i.rand.yield } };
x = a.asStream;
x.nextN(8);
::

This is useful for using ListComprehensions in Patterns:

code::
Pbind(\degree, p {:[x, y].postln, x<-(0..10), y<-(0..10), (x + y).isPrime }, \dur, 0.3).play;
::

subsection:: Bela

method:: belaScope
Like link::Classes/Function#-scope::, plays and scopes this Function, but using Bela's Oscilloscope (see link::Classes/BelaScope:: for required setup).
This Function's output bus is monitored right after the generated Synth.

argument:: scopeChannel
Bela's oscilloscope channel to start scoping on. This has to be a non-negative number, and can't be changed after scoping starts.

argument:: target
A Node, Server, or Nil. A Server will be converted to the default group of that server. Nil will be converted to the default group of the default Server.
note:: this Function's play target needs to be on a Server that supports BelaScope::

argument:: numChannels
Number of channels to scope, from this Function's output channels. Defaults to this Function's output channels.

argument:: outbus
argument:: fadeTime
argument:: addAction
argument:: args
See link::Classes/Function#-scope::

returns:: Like link::Classes/Function#-play::, a Synth playing this function. When that synth is freed, so will be its monitor.

examples::

subsection::Exception Handling

code::
// no exception handler
value { 8.zorg; \didnt_continue.postln; }

try { 8.zorg } {|error| error.postln; \cleanup.postln; }; \continued.postln;

protect { 8.zorg } {|error| error.postln; }; \didnt_continue.postln;
::

code::
try { 123.postln; 456.throw; 789.postln } {|error| [\catch, error].postln };

try { 123.postln; 789.postln } {|error| [\catch, error].postln };

try { 123.postln; nil.throw; 789.postln } {|error| [\catch, error].postln };

protect { 123.postln; 456.throw; 789.postln } {|error| [\onExit, error].postln };

protect { 123.postln; 789.postln } {|error| [\onExit, error].postln };

(
try {
	protect { 123.postln; 456.throw; 789.postln } {|error| [\onExit, error].postln };
} {|error| [\catch, error].postln };
)

value { 123.postln; 456.throw; 789.postln }

value { 123.postln; Error("what happened?").throw; 789.postln }
::

code::
(
a = [\aaa, \bbb, \ccc, \ddd];
a[1].postln;
a[\x].postln;
a[2].postln;
)

(
try {
	a = [\aaa, \bbb, \ccc, \ddd];
	a[1].postln;
	a[\x].postln;
	a[2].postln;
} {|error| \caught.postln; error.dump }
)

(
try {
	a = [\aaa, \bbb, \ccc, \ddd];
	a[1].postln;
	a[\x].postln;
	a[2].postln;
} {|error| \caught.postln; error.dump; error.throw }
)

(
protect {
	a = [\aaa, \bbb, \ccc, \ddd];
	a[1].postln;
	a[\x].postln;
	a[2].postln;
} {|error| \caught.postln; error.dump }
)
::