tinyfunk

The tiniest of functional libraries.

Documentation

• Motivation
• Caveat Emptor
• API
• Dependents

Motivation

Most popular functional libraries, like Ramda, are quite large. To use them in a frontend project, it's common to import only the bits you need (to keep the bundle size down) like this:

const assoc = require('ramda/src/assoc')
const map   = require('ramda/src/map')
const merge = require('ramda/src/merge')
// etc.

But multiple require statements take up extra space, since many js code compression tools (including uglify-js) don't mangle strings. The AST overhead required to bundle the various modules adds an additional size penalty, not to mention the extra compute time to parse the twisted flow of the AST, which adds to a longer perceived load time.

tinyfunk takes a different approach. It exports a single module to minimize AST overhead and obviate the repeated require statements. You just destructure the functions you need in one go:

const { assoc, map, merge } = require('tinyfunk')

Where possible, it also composes more complex functions by reusing basic ones. A good example is compose, implemented as so:

const compose = unapply(flip(reduceRight(thrush)))

With uglify-es, this mangles down to the following. I doubt you'll find a smaller implementation.

const I=r(b(B(E)))

Caveat emptor

In an effort to keep tinyfunk lean and mean - and most of all, tiny - I've taken a few shortcuts.

• None of the exported functions perform type-checking of arguments.
  If type-checking is a runtime debug tool you tend to lean on, then you'll need to look elsewhere.

• Many of the function combinators only support unary functions.
  This includes compose, converge, juxt, pipe, thrush, etc. Unary functions are easily composable, readily portable, and so much simpler (ie: tinier) to support.

• Unlike other popular FP libraries, each exported function only has a single job.
  For example, Ramda's map supports "mapping" over functors, objects, and even functions. In tinyfunk, those various jobs are supported instead by map, mapObj, and compose respectfully. So be sure to use the right tool for the job at hand.

API

If you've lived with FP long enough, you are likely familiar with most of the functions listed below. So I've opted to leave out the lengthy descriptions and include only the signatures. I'll be adding more functions as I need them, but if you see any of your favorites missing, just post an issue and I'll be sure to consider it.

Function	Signature
add	Number -> Number -> Number
append	a -> [a] -> [a]
apply	(a... -> b) -> [a] -> b
assoc	k -> v -> { k: v } -> { k: v }
assocPath	[k] -> v -> { k: v } -> { k: v }
call	(a... -> b) -> a... -> b
compose	((y -> z), ..., (a -> b)) -> a -> z
composeP	((y -> Promise z), ..., (a -> Promise b)) -> a -> Promise z
concat	Semigroup a => a -> a -> a
cond	[[(a -> Boolean), (a -> b)]] -> a -> b
constant	a -> () -> a
converge	(b... -> c) -> [(a -> b)] -> a -> c
curry	((a, b, ...) -> z) -> a -> b -> ... -> z
curryN	Number -> ((a, b, ...) -> z) -> a -> b -> ... -> z
defaultTo	a -> a -> a
dissoc	k -> { k: v } -> { k: v }
dissocPath	[k] -> { k: v } -> { k: v }
either	(a -> Boolean) -> (a -> Boolean) -> (a -> Boolean)
evolve	{ k: (v -> v) } -> { k: v } -> { k: v }
filter	(a -> Boolean) -> [a] -> [a]
find	(a -> Boolean) -> [a] -> a
flatten	[a] -> [b]
flip	(a -> b -> c) -> (b -> a -> c)
head	[a] -> a
identity	a -> a
ifElse	(a -> Boolean) -> (a -> b) -> (a -> b) -> (a -> b)
indexBy	(v -> k) -> [v] -> { k: v }
init	[a] -> [a]
is	Constructor -> a -> Boolean
join	String -> [a] -> String
juxt	[(a -> b)] -> a -> [b]
keys	{ k: v } -> [k]
last	[a] -> a
length	[a] -> Number
map	Functor f => (a -> b) -> f a -> f b
mapObj	(v -> k -> v) -> { k: v } -> { k: v }
match	RegExp -> String -> [String]
merge	{ k: v } -> { k: v } -> { k: v }
multiply	Number -> Number -> Number
nAry	Number -> (a... -> b) -> (a... -> b)
not	a -> a
objOf	k -> v -> { k: v }
omit	[k] -> { k: v } -> { k: v }
partial	(a... -> b) -> [a] -> a... -> b
partialRight	(a... -> b) -> [a] -> a... -> b
path	[k] -> { k: v } -> v
pathEq	[k] -> v -> { k: v } -> Boolean
pick	[k] -> { k: v } -> { k: v }
pipe	((a -> b), ..., (y -> z)) -> a -> z
pipeP	((a -> Promise b), ..., (y -> Promise z)) -> a -> Promise z
pluck	k -> [{ k: v }] -> [v]
prepend	a -> [a] -> [a]
prop	k -> { k: v } -> v
propEq	k -> v -> { k: v } -> Boolean
props	[k] -> { k: v } -> [v]
reduce	Foldable f => (b -> a -> b) -> b -> f a -> b
reduceObj	(a -> v -> k -> a) -> a -> { k: v } -> a
reduceP	(b -> a -> Promise b) -> b -> [a] -> Promise b
reduceRight	Foldable f => (b -> a -> b) -> b -> f a -> b
reduceRightP	(b -> a -> Promise b) -> b -> [a] -> Promise b
replace	RegExp -> String -> String -> String
slice	Number -> Number -> [a] -> [a]
sort	((a, a) -> Number) -> [a] -> [a]
sortBy	Ord b => (a -> b) -> [a] -> [a]
split	RegExp -> String -> [String]
tail	[a] -> [a]
take	Number -> [a] -> [a]
tap	(a -> b) -> a -> a
test	RegExp -> String -> Boolean
then	(a -> Promise b) -> a -> Promise b
thrush	a -> (a -> b) -> b
unapply	([a] -> b) -> a... -> b
unary	(a... -> b) -> (a -> b)
uniqBy	(a -> String) -> [a] -> [a]
unit	a -> ()
unless	(a -> Boolean) -> (a -> a) -> a -> a
useWith	(b... -> c) -> [(a -> b)] -> a... -> c
values	{ k: v } -> [v]
when	(a -> Boolean) -> (a -> a) -> a -> a
zipObj	[k] -> [v] -> { k: v }

Dependents

I use tinyfunk everyday to make cool things. Maybe you do, too. If so, let me know with a PR, and we can add your cool things to this list.

• puddles - Tiny vdom app framework. Pure Redux. No boilerplate. - Built with ❤️ on tinyfunk.