@playframe/statue
v1.0.4
Published
Art of State Management
Downloads
1
Readme
Statue
0.4 kb Art of functional State
Installation
npm install --save @playframe/statue
Description
Statue is a Redux like functional state manegement library that lets you define and access actions right inside of your state. So you could describe a deeply nested state tree with nested actions in one simple object.
To update state, action could return a new object that has one or more properties. This will produce a new state with those properties updated.
Another option is to mutate a state object passed to your function. Mutation will be detected and this will produce a new state as well. Returned value is ignored in this case. If you are writing a performance demanding reducer, please use this mutation trick and define the hottest property as the first one in your initial state for faster checking.
If you don't want to update state, please make sure you action returns something falsy or the same state that was passed to it
To work with state directly you can do
// to update the state
state._(updated)
// to read the latest state
state = state._()
// to turn a function into a reducer
state._.new_action = state._( (x, state)=> state.x = x )
Statue also cares not to send too many updates from child to parent branches
Usage
import statue from '@playframe/statue'
const state = statue({
i: 0,
_: {
increment: (x, state)=> i: state.i + 1,
incrementBy: (n)=>(event, state)=> state.i += n // mutating
},
subState: {
i: 0,
_: {
increment: (x, state)=> i: state.i + 1
}
}
},
setTimeout,
(newState)=> {
console.log('Counter is:', newState.i)
console.log('Subcounter is:', newState.subState.i)
}
)
for(let j = 0; j < 100; j++){
state.subState._.increment()
state.subState._.increment()
state._.increment()
}
// Logs only once
//> Counter is: 100
//> Subcounter is: 200
// Will increment by 10 and log in console on every click
$('button').click( state._.incrementBy(10) )
Annotated Source
Caching static functions from Object
{assign, create, keys} = Object
is_function = (f)=>
typeof f is 'function'
Let's define a function that takes a takes state_actions
as an initial state and its reducers (actions)
defined under _
(underscore) property.
Our function also takes level_up
or 'subscribe' function
that will be called when state updates. Most sertainly we don't
want to update the whole state tree every time nested leaf updates,
so we also pass a delayed
function that will help debounce
parent update
module.exports = statue = (state_actions, delayed, level_up)=>
### state._.actions
]_[.state._.actions
[_] ]_[.
[_] [_]
###
actions = state_actions._
_state = state_actions # _closure
_scheduled = false # _closure
_nested = reset = => _nested = reset # reset _nested
# delaying child state updates
delay_nested = (f)=>
_nested = do (_nested)=>=> do _nested; do f; return
do schedule
return
# recursive statue if there nested actions
for k, v of state_actions when v and v._
_state[k] = statue v, delay_nested, do (k)=>(sub_state)=> # closure for k
_state[k] = sub_state # executes as _nested
# saving new state in closure
save_state = (state)=>
do schedule
_state = state
schedule = =>
unless _scheduled
_scheduled = true
# lazy parent update
delayed update_parent
return
update_parent = =>
_scheduled = false
proto_state = _state
_state = create null
# merging prototype chain state
_state[k] = proto_state[k] for k of proto_state
do _nested
level_up _state
return
This function is a little overloaded, it's a getter/setter but
also is a function wrapper.
You can access it like this yourState._(updated)
.
The wrapper will return a twin of your function
that calls yours and passes a copy of latest state
as the second argument. State will update, if the copy is mutated
or a new object returned. If your function
returns a new function, it will be wrapped in the same manner.
So we support state updates for deeply curried functions.
# curry down or make state
_state._ = update_state = (arg)=>
if is_function arg
inject_state arg
else
if arg
save_state assign create(_state), arg
else
_state
inject_state
is a higher order function that does the magic.
It produces a cheap clone if current state by setting current
state as its prototype. Please note that such a clone has no
own properties, and all property accees falls back to its
prototype. If we mutate such a clone, new properties are
easily detected
inject_state = (f)=>(x)=>
# _state as prototype of cloned
cloned = create _state
y = f x, cloned
for own k of cloned
# mutation detected
mutated = true
save_state cloned
break
if is_function y
# recursevely currying down
inject_state y
else
if y and not mutated and y isnt cloned and not y.then # not promise
save_state assign cloned, y
y
Now let's wrap all of your actions and set under the _
property.
That's it! You new state machine is ready!
# bind actions to state
for k, action of actions
update_state[k] = inject_state action
_state