README.md

Puff

Puff is a library for Point Free or Function Level programming in Javascript, inspired by the terseness and productivity (for single developers) of languages like J, APL, and K/Q.

As a person who writes software, I have at least two modes: in one, I am a team member writing software which is supposed to be maintainable and understandable by other developers. As (for instance) a generative artist or data analyst I am interested in quickly sketching out high level pipelines or programs which I can depend on being correct.

Puff is designed for the latter use case: it allows the Javascript programmer to express things with great conciseness.

For instance:

 function normalizeToHeight(s,h){
	 var minVal = s.reduce(min2);
	 var maxVal = s.reduce(max2);
	 return map(r(p_(minus,minVal), 
		          p_(div, maxVal-minVal), 
		          p_(times, h)), s);
 }

Uses the reverse function composition combinator (r) to compose three partially applied functions (p_ partially applies on the right, _p on the left) to normalize the signal s so that it is between zero and h.

The above example mixes Javascript style with function level programming (allowing a standard function definition to bind s and h). We can choose how deep down the function level programming hole we want to go with puff:

var normalizeToHeight = r(a,clo_({s:first,
                                 h:second,
								 minVal:r(first, p_(reduce,min)),
								 maxVal:r(first, p_(reduce,max)})),
                          au(map, au(r,
						             au_(minus, p_(ix, 'minVal')),
									 au_(div,au_(minux, p_(ix, 'maxVal'), p_(ix, minVal))),
									 au_(times, p_(ix, 'h'))),
								 p_(ix, 's'));

This is a bit much, though. It is instructive, nevertheless, to read the above:

r introduces a composition - each argument is a function, each of which is applied in order from left to right.

a returns all of its arguments as a single array.

clo_ (cleaveObject_) returns a function waiting for a single value and applies the functions at each key in the object provided to that value and returns the object constructed by assigning each result to its key.

clo_ is just c_(clo), eg, clo curried rightways.

au (augment) takes a function and returns a new function onto which argument wise composition has occured. Eg:

au(function(a,b){return a+b}, first, second)

Is equivalent to the function:

function(v){ return first(v)+second(v) };

By the above definition we can see that au on r above gives us the composition of substracting that value stored at the minVal index, dividing that by the difference between that stored at maxVal and minVal.

map is finally applied to the value at key s.

If this seems kind of insane to you, well, that is function level programming. Several new combinators would make the above more succinct: aumap:_p(au,map), aur:_p(au,r) or even better: aurap:_p(ap,_p(au,r)).