Skip to content

πŸ” An Opinionated and Painless React n Redux promise wrapper

License

Notifications You must be signed in to change notification settings

untra/naiveasync

Folders and files

NameName
Last commit message
Last commit date

Latest commit

Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 

Repository files navigation

πŸ” NaiveAsync

an opinionated and painless React n Redux promise wrapper

Simultaneously master redux and never have to write it ever again

NaiveAsync is a variety of utilities for cleanly turning promises into state for react components, managed in a redux store. It is a straightforward React functional module that can be used to quickly turn an asynchronous function into a managed and cached lifecycle object that can represent components in various states.

Real quick: an async function can be specified with the async function keyword, or can be understand to be a function that returns a promise aka () => Promise. Typescript makes this abundantly clear.

NaiveAsync will autogenerate your redux selectors, dispatch operations, and provide a managed lifecycle object you can control around your async operations for use in react components.

The AsyncLifecycle turns an async function into a stateful object to render child components with, redux action creators, selectors, and reducers to manage the promise state in redux, with a bunch of other tools for combining operation wiring. It even provides test shapes and mock interfaces.

basically its a swiss army knife that does all your redux for you and makes your storybook better

To invoke your promise, you dispatch either .call() or .sync() to your redux store. The difference is subtle:

  • use .call() to call the promise and reset the state of the lifecycle immediately. Good for when you want the current data invalidated ASAP before any data is returned. For example: create, update, delete operations, tests and dynamic operations.

  • use .sync() to call the promise and retain the existing state of the lifecycle until data is returned. Note that as a convenience, subsequent sync invocations without parameters will reuse the last specified parameters. Good for: get, indexes, subscriptions, and searches, idempotent operations

Usage

NaiveAsync is a variety of utilities. At its core are two main tools though:

  • asyncLifecycle - which turns a () => Promise into the swiss army knife of managed redux juicy goodness
  • <NaiveAsync> - all that as an exotic react component

asyncLifecycle

This function takes in an async function and an identifier that is used in the redux store to namespace your promise state. As the promise executes, the state of the promise is recorded in a static object that can be selected efficiently:

If you're interested in what it does, it'll be easier just to point you to the code.

NaiveAsync

import * as ReactDOM from "react-dom";
import { Provider } from "redux";
import { applyMiddleware, createStore } from "redux";
import { NaiveAsync, naiveAsyncMiddleware, naiveAsyncReducer } from '@untra/naiveasync'
// the naiveAsyncReducer maintains the redux state
// the naiveAsyncMiddleware employs rxjs observables to fulfill promises
const store = createStore(naiveAsyncReducer, applyMiddleware(naiveAsyncMiddleware))

// an async function is a function that returns a promise.
const asyncOperation = (val: number) => new Promise((resolve) => {
  const timeMS = Math.random() * 4000
  setTimeout(() => resolve(val), timeMS)
})

// supply the created store into your redux provider
// use the <NaiveAsync>(state, call) => react component and callback
// to render your asynchronous state with ease and splendor
ReactDOM.render(
<Provider store={store}>
    <NaiveAsync operation={asyncOperation} id={"example"}>(state, call) => (
        state.data ? <h2>{state.data}</h2> :
        state.status === 'inflight' ? <h2>loading...</h2> :
        <button onclick={call()}>call</button>
    )
    </NaiveAsync>
</Provider>
)

Design

NaiveAsync uses react 16.8.5 hooks to create an asynchronous experience you could take home to your mother.

Promises are a powerful tool in javascript, and a wrapper to abstract its most common uses into a simple react component that just works is the goal here.

Some Terminology:

  • an AsyncOperation<D,P> is a function that takes (P) and returns a Promise<D>
  • an AsyncState is an object of type
{
  status: '' | 'inflight' | 'error' | 'done'
  error: string
  params: <P extends {}>
  data: null|D
}

Recommended usage with REST APIs

  • use .OnData to dispatch a .sync indexing calls on create, update and destroy
  • use .debounce on search calls, to ensure that the underlying operation is not repeated
  • use a 2 - 6 seconds .subscribe for live dashboards and gentle api usage.
  • use a (very low > 1000 ms) .throttle on index calls, so multiple resources that need it but request slow can receive it on the same batch
  • sensitive calls that should not throw an error (eg. login / auth) can use .onError to dispatch emergency logout
  • calls to the api whose response is not likely to change can use .memoize(true)
  • set a lifecycle dataDepends to pause execution until data has come back on separate lifecycles
  • subscribed calls that should cancel awaiting lifecycles should implement AbortController for the promise if possible.
  • testing utilities pair nicely with storybook
  • best served with typescript

Recommended usage with Testing

  • use lifecycle .meta() to get the AsyncMeta, a snapshot printout of the metacache for this lifecycle, useful in testing
  • use mockInitialAsyncState, mockInflightAsyncState, mockErrorAsyncState, mockDoneAsyncState to represent async states in mocks and storybook scenes
  • dispatch the lifecycle .assign(state) action to assign a specific state to the lifecycle. this is typically frowned upon in redux philosophy, but is really helpful in mocking state
  • when testing within async functions, use await lifecycle.awaitResolve() to pause test execution until the async operation next resolves. similarly use await lifecycle.awaitReject() to test rejection.
  • lifecycle's can be passed options, and optionally passed { traceDispatch: true } to add a stacktrace to dispatched actions, and to trace invocations of a lifecycle.
  • call .invalidate() after you mock action dispatching functions on a lifecycle to keep the lifecycle fresh.

1.0.0 feature wishlist:

  • swap placement of P and D, rename the dang thing
  • rename the AsyncState type
  • .timeout() will stop the async function and error after a specified timeout
  • .subscribe() retries the request on a given interval
  • .onData((data? : D, dispatch? : Dispatch<AnyAction>) => void) data callback with dispatch function
  • .onError((error? : string, dispatch? : Dispatch<AnyAction>) => void) error callback with dispatch
  • .memoized(enabled? : boolean = true) keeps a record of inputs and their outputs, and returns the cached results
  • .exponentialErrorRetry(enabled? : boolean = true) if the request errors, retries the request if it fails a few seconds from now, following an exponential backoff strategy
  • lifecycle .meta display meta information about the selector eg:
    • consecutive data count
    • consecutive error count
    • time inflight : number
    • timeout number
    • debounce number
    • throttle number
    • error retry bool
    • data retry bool
  • AbortController support
  • test support for immutablejs
  • test support as observable / generator

How it works

NaiveAsync maintains two caches of lifecycles and their meta properties (three if you include the redux state). The asyncLifecycle(id, async () => { ... }) first argument is an id that will be checked against a cache. As a result, that function may return a reference to an existing lifecycle. A lifecycle can reset itself by calling .invalidate().

Similar but worse

To achieve the similar goals as what react + redux + naiveasync can provide, heres a starting point for your other framework or whatever

Copyright

Copyright (c) Samuel Volin 2021. License: MIT