Extending re-ducks proposal

gui commited 6 years ago · react redux

Take a look at demo or repository.

Before starting you may want to check both ducks and re-ducks proposals. Although both proposals suits me very well and improved a lot my workflow, I was still missing some more definitions about how to handle operations properly. Also, while handling such operations (which were basically API calls most of time) I noticed that my apps was repeating the same API request in a short time.

Not just that, but I also noticed I was always following same steps:

So, I would like to extend both original ideas (which are great) to support Operation and Cached Responses.

Original ideas

Duck rules

A module...

  1. MUST export default a function called reducer()
  2. MUST export its action creators as functions
  3. MUST have action types in the form npm-module-or-app/reducer/ACTION_TYPE
  4. MAY export its action types as UPPER_SNAKE_CASE, if an external reducer needs to listen for them, or if it is a published reusable library

Re-ducks rules

A duck folder...

  1. MUST contain the entire logic for handling only ONE concept in your app, ex: product, cart, session, etc.
  2. MUST have an index.js file that exports according to the original duck rules.
  3. MUST keep code with similar purpose in the same file, ex: reducers, selectors, actions, etc.
  4. MUST contain the tests related to the duck.

Cache Re-ducks implementation

Types

Will follow same re-ducks recipe, but now will have a new rule for it:

In other words: Requests that can be successful or failed should have same single constant type.

So, you will avoid to have types like:

const FETCH_DATA   = "app/data/FETCH_DATA";
const RECEIVE_DATA = "app/data/RECEIVE_DATA";
const FAILED_DATA  = "app/data/FAILED_DATA";

To have just:

const FETCH_DATA = "app/data/FETCH_DATA";

Actions

Callback actions should always receive result (whether success or fail), data (that can be an error or succeed info) and as many additional parameters you want.

import types from './types';

const fetchData = () => {
    return {
        type: types.FETCH_DATA        
    }
};

const receiveFetchData = (result, data) => {
    if (result === "success") {
        return {
            type: types.FETCH_DATA,
            result,
            data
        }
    }

    return {
        type: types.FETCH_DATA,
        result,
        errors: data
    }
};

export default {
    fetchData,
    receiveFetchData
};

Operations

Now you get a basic operation class (that should be extended) with some methods that may be overridden.

MethodDescription
constructorExpects at least two actions for request and callback respectively
dispatchInvokes operation itself. Probably you don't want to change it.
getIdDefines that a request is unique. By default, it generates a hash based on all property values.
createApiServiceGenerates whatever you use to perform requests.
onRequestCalled before doing first request. By default dispatches request action.
onGetResponseFirst request callback. By default returns response.data.
getSucceedDataReceive data from onGetResponse and prepare data to onSucceed. By default just pass same arg received from onGetResponse.
onSucceedCalled when request is succeed. By default dispatches callback action with "success" and "data" as parameters.
onFailedCalled when request fails. By default dispatches callback action with "fail" and "errors" as parameters. Uses an empty generic error handler to be customized if you want to
getApiPromise*Handles api call and returns a promisse. No defaults, must be implemented.

Basic usage example

import actions from './actions';

class FetchProductOperation extends Operation {
    constructor(id) {
        super(actions.fetchProductDetails, actions.receiveProductDetails);
        this.id = id;
    }

    getApiPromise(api) {
        const url = "products/" + this.id;
        return api.get(url);
    }
}

const fetchProduct = (id) => new FetchProductOperation(id).dispatch();

export default {
    fetchProduct
};

Why a class?

Just because is simple to be overridden and easily understood. You don't need to change every method, just those that are specific for some case you're handling.

Cache MUSTS

A cache MUST...

  1. Avoid repeating same requests in a too small lenght of time.
  2. Have it's timeout lenght set individually for each operation, be customizable, and easily enabled/disabled.
  3. Be invisible to operation and user. So it doesn't mess up with our logic and can be added or removed as you will.
  4. When get stale cache a new result alongside request time.

So you start with following function:

import { operations, selectors } from '../core-duck/duck';

const cacheTimeoutMultiplier = 1000 * 60; //Convert to minutes

const cacheOperation = (operation, timeout) => (dispatch, getState) => {
    timeout = timeout * cacheTimeoutMultiplier;
    const cache = selectors.getCacheFromId(getState(), operation.getId());
    const isStale = cache ? (Date.now() - cache.time) > timeout : true;

    if (isStale) {
        // Do API request as usual, but when succeed, cache result
        operation.onGetResponse = (response) => {
            dispatch(cacheResponse(operation.getId(), Date.now(), response.data));
            return response.data;
        };
    }    
    else {
        // Don't do any request, use cached result
        operation.getApiPromise = () => {
            return Promise.resolve(cache.result);
        };
        operation.onGetResponse = (data) => {
            return data;
        };
    }

    return operation.dispatch()(dispatch, getState);
};

export default cacheOperation;

To be applied to any already existing operation, like:

import actions from './actions';
import cache from '../cacheOperation';

class FetchProductOperation extends Operation {
    constructor(id) {
        super(actions.fetchProductDetails, actions.receiveProductDetails);
        this.id = id;
    }

    getApiPromise(api) {
        const url = "products/" + this.id;
        return api.get(url);
    }
}

const fetchProduct = (id, timeout = 5) => cache(new FetchProductOperation(id), timeout);

export default {
    fetchProduct
};

In this example we've set any repeated request to same product id inside 5 minutes to be cached meeting item 1.

We can easily pass a 0 for fetchProduct, disabling cache if we will or even passing a minor timeout. We can change cacheTimeoutMultiplier if we want to work with seconds or other thing. We can even create another fetchProduct with no cache config at all meeting our item 2.

It works exactly as any other operation, nor user nor your app needs to know that result was cached. In the end it's just a resolved promisse with succeed result. Meeting item 3 :) .

There's no need to explicitly cache it in our operation, our cache func handles it easily. You just need to setup a basic core duck to handle this kind of stuff. Item 4 met!

Why caching?

This may change from app to app, only you can say if it's really needed. Let's imagine that your app lists some patients, if you perform a GET request to /patients at 15:00:00, and at 15:05:30 you perform it again just because your user come back to this screen from another one. What's the chances to something have changed in just 05:30 minutes? If nothing has changed, your user waits for nothing and your server works for nothing. Of course, your bussiness rules and environment will give a final word on this.

For such reason, I propose something highly customizable and easy to turn on or off. Maybe you can't even think about caching patients, but there's still a medications stock list that are only updated once a few hours. Maybe you can cache it for 15 minutes? Who knows? It' up to you. Remember your user doesn't need to be stuck to your predefined timeout, you can always set a refresh button to trigger same operation passing a 0 timeout param.

Example app

You can check example folder with an app built using material-ui + cache re-ducks structure.

App basically retrieves a temperature from a location using OpenWeather api, caching requests to same location in a lenght of 2 minutes. Cache can be disabled and timeout changed by user, also it was customized to extend some functions and meet some needs for this project.

Conclusion

Of course this project still need to grow, also I could improve by adding tests and more. Sadly I'm kind out of time lately, so I tried hard to create at least a good example showing how to implement such proposal. I'm always open to suggestions, feel free to contribute :) .

I thank God for the opportunity to showcase my proposal. Hope someone find it useful.

  • LinkedIn
  • Tumblr
  • Reddit
  • Google+
  • Pinterest
  • Pocket