Develop Reference

Is it safe to call sagaMiddleware.run multiple times?

I'm using redux and redux-saga in an application to manage state and asynchronous actions. In order to make my life easier, I wrote a class that acts essentially as a saga manager, with a method that "registers" a saga. This register method forks the new saga and combines it with all other registered sagas using redux-saga/effects/all:
class SagasManager {
public registerSaga = (saga: any) => {
this._sagas.push(fork(saga));
this._combined = all(this._sagas);
}
}
This class is then used by my store to get the _combined saga, supposedly after all sagas are registered:
const store = Redux.createStore(
reducer,
initialState,
compose(Redux.applyMiddleware(sagaMiddleware, otherMiddleware)),
);
sagaMiddleware.run(sagasManager.getSaga());
However, I ran into the problem that depending on circumstances (like import order), this doesn't always work as intended. What was happening was that some of the sagas weren't getting registered before the call to sagaMiddleware.run.
I worked around this by providing a callback on SagasManager:
class SagasManager {
public registerSaga = (saga: any) => {
this._sagas.push(fork(saga));
this._combined = all(this._sagas);
this.onSagaRegister();
}
}
And then the store code can use this as
sagasManager.onSagaRegister = () => sagaMiddleware.run(sagasManager.getSaga());
This seems to work, but I can't find in the docs whether this is safe. I did see that .run returns a Task, which has methods for canceling and the like, but since my problem is only in that awkward time between when the store is constructed and the application is rendered I don't that would be an issue.
Can anyone explain whether this is safe, and if not what a better solution would be?

It may depend on what you mean by "safe". What exactly do you mean by that in this case?
First, here's the source of runSaga itself, and where it gets used by the saga middleware.
Looking inside runSaga, I see:
export function runSaga(options, saga, ...args) {
const iterator = saga(...args)
// skip a bunch of code
const env = {
stdChannel: channel,
dispatch: wrapSagaDispatch(dispatch),
getState,
sagaMonitor,
logError,
onError,
finalizeRunEffect,
}
const task = proc(env, iterator, context, effectId, getMetaInfo(saga), null)
if (sagaMonitor) {
sagaMonitor.effectResolved(effectId, task)
}
return task
}
What I'm getting out of that is that nothing "destructive" will happen when you call runSaga(mySagaFunction). However, if you call runSaga() with the same saga function multiple times, it seems like you'll probably have multiple copies of that saga running, which could result in behavior your app doesn't want.
You may want to try experimenting with this. For example, what happens if you have a counter app, and do this?
function* doIncrement() {
yield take("DO_INCREMENT");
put({type : "INCREMENT"});
}
sagaMiddleware.runSaga(doIncrement);
sagaMiddleware.runSaga(doIncrement);
store.dispatch({type : "DO_INCREMENT"});
console.log(store.getState().counter);
// what's the value?
My guess is that the counter would be 2, because both copies of doIncrement would have responded.
If that sort of behavior is a concern, then you probably want to make sure that prior sagas are canceled.
I actually ran across a recipe for canceling sagas during hot-reloading a while back, and included a version of that in a gist for my own usage. You might want to refer to that for ideas.

simplify redux with generic action & reducer

In React-Redux project, people usually create multiple actions & reducers for each connected component. However, this creates a lot of code for simple data updates.
Is it a good practice to use a single generic action & reducer to encapsulate all data changes, in order to simplify and fasten app development.
What would be the disadvantages or performance loss using this method. Because I see no significant tradeoff, and it makes development much easier, and we can put all of them in a single file! Example of such architecture:
// Say we're in user.js, User page
// state
var initialState = {};
// generic action --> we only need to write ONE DISPATCHER
function setState(obj){
Store.dispatch({ type: 'SET_USER', data: obj });
}
// generic reducer --> we only need to write ONE ACTION REDUCER
function userReducer = function(state = initialState, action){
switch (action.type) {
case 'SET_USER': return { ...state, ...action.data };
default: return state;
}
};
// define component
var User = React.createClass({
render: function(){
// Here's the magic...
// We can just call the generic setState() to update any data.
// No need to create separate dispatchers and reducers,
// thus greatly simplifying and fasten app development.
return [
<div onClick={() => setState({ someField: 1 })}/>,
<div onClick={() => setState({ someOtherField: 2, randomField: 3 })}/>,
<div onClick={() => setState({ orJustAnything: [1,2,3] })}/>
]
}
});
// register component for data update
function mapStateToProps(state){
return { ...state.user };
}
export default connect(mapStateToProps)(User);
Edit
So the typical Redux architecture suggests creating:
Centralized files with all the actions
Centralized files with all the reducers
Question is, why a 2-step process? Here's another architectural suggestion:
Create 1 set of files containing all the setXField() that handle all the data changes. And other components simply use them to trigger changes. Easy. Example:
/** UserAPI.js
* Containing all methods for User.
* Other components can just call them.
*/
// state
var initialState = {};
// generic action
function setState(obj){
Store.dispatch({ type: 'SET_USER', data: obj });
}
// generic reducer
function userReducer = function(state = initialState, action){
switch (action.type) {
case 'SET_USER': return { ...state, ...action.data };
default: return state;
}
};
// API that we export
let UserAPI = {};
// set user name
UserAPI.setName = function(name){
$.post('/user/name', { name }, function({ ajaxSuccess }){
if (ajaxSuccess) setState({ name });
});
};
// set user picture URL
UserAPI.setPicture = function(url){
$.post('/user/picture', { url }, function({ ajaxSuccess }){
if (ajaxSuccess) setState({ url });
});
};
// logout, clear user
UserAPI.logout = function(){
$.post('/logout', {}, function(){
setState(initialState);
});
};
// Etc, you got the idea...
// Moreover, you can add a bunch of other User related methods,
// like some helper methods unrelated to Redux, or Ajax getters.
// Now you have everything related to User available in a single file!
// It becomes much easier to read through and understand.
// Finally, you can export a single UserAPI object, so other
// components only need to import it once.
export default UserAPI
Please read through the comments in the code section above.
Now instead of having a bunch of actions/dispatchers/reducers. You have 1 file encapsulating everything needed for the User concept. Why is it a bad practice? IMO, it makes programmer's life much easier, and other programmers can just read through the file from top to bottom to understand the business logic, they don't need to switch back and forth between action/reducer files. Heck, even redux-thunk isn't needed! And you can even test the functions one by one as well. So testability is not lost.

Firstly, instead of calling store.dispatch in your action creator, it should return an object (action) instead, which simplifies testing and enables server rendering.
const setState = (obj) => ({
type: 'SET_USER',
data: obj
})
onClick={() => this.props.setState(...)}
// bind the action creator to the dispatcher
connect(mapStateToProps, { setState })(User)
You should also use ES6 class instead of React.createClass.
Back to the topic, a more specialised action creator would be something like:
const setSomeField = value => ({
type: 'SET_SOME_FIELD',
value,
});
...
case 'SET_SOME_FIELD':
return { ...state, someField: action.value };
Advantages of this approach over your generic one
1. Higher reusability
If someField is set in multiple places, it's cleaner to call setSomeField(someValue) than setState({ someField: someValue })}.
2. Higher testability
You can easily test setSomeField to make sure it's correctly altering only the related state.
With the generic setState, you could test for setState({ someField: someValue })} too, but there's no direct guarantee that all your code will call it correctly.
Eg. someone in your team might make a typo and call setState({ someFeild: someValue })} instead.
Conclusion
The disadvantages are not exactly significant, so it's perfectly fine to use the generic action creator to reduce the number of specialised action creators if you believe it's worth the trade-off for your project.
EDIT
Regarding your suggestion to put reducers and actions in the same file: generally it's preferred to keep them in separate files for modularity; this is a general principle that is not unique to React.
You can however put related reducer and action files in the same folder, which might be better/worse depending on your project requirements. See this and this for some background.
You would also need to export userReducer for your root reducer, unless you are using multiple stores which is generally not recommended.

I mostly use redux to cache API responses mostly, here are few cases where i thought it is limited.
1) What if i'm calling different API's which has the same KEY but goes to a different Object?
2) How can I take care if the data is a stream from a socket ? Do i need to iterate the object to get the type(as the type will be in the header and response in the payload) or ask my backend resource to send it with a certain schema.
3) This also fails for api's if we are using some third party vendor where we have no control of the output we get.
It's always good to have control on what data going where.In apps which are very big something like a network monitoring application we might end up overwriting the data if we have same KEY and JavaScript being loosed typed may end this to a lot weird way this only works for few cases where we have complete control on the data which is very few some thing like this application.

Okay i'm just gonna write my own answer:
when using redux ask yourself these two questions:
Do I need access to the data across multiple components?
Are those components on a different node tree? What I mean is it isn't a child component.
If your answer is yes then use redux for these data as you can easily pass those data to your components via connect() API which in term makes them containers.
At times if you find yourself the need to pass data to a parent component, then you need to reconsider where your state lives. There is a thing called Lifting the State Up.
If your data only matters to your component, then you should only use setState to keep your scope tight. Example:
class MyComponent extends Component {
constructor() {
super()
this.state={ name: 'anonymous' }
}
render() {
const { name } = this.state
return (<div>
My name is { name }.
<button onClick={()=>this.setState({ name: 'John Doe' })}>show name</button>
</div>)
}
}
Also remember to maintain unidirectional data flow of data. Don't just connect a component to redux store if in the first place the data is already accessible by its parent component like this:
<ChildComponent yourdata={yourdata} />
If you need to change a parent's state from a child just pass the context of a function to the logic of your child component. Example:
In parent component
updateName(name) {
this.setState({ name })
}
render() {
return(<div><ChildComponent onChange={::this.updateName} /></div>)
}
In child component
<button onClick={()=>this.props.onChange('John Doe')}
Here is a good article about this.
Just practice and everything will start to make sense once you know how to properly abstract your app to separate concerns. On these matter composition vs ihhertitance and thinking in react are a very good read.

I started writing a package to make it easier and more generic. Also to improve performance. It's still in its early stages (38% coverage). Here's a little snippet (if you can use new ES6 features) however there is also alternatives.
import { create_store } from 'redux';
import { create_reducer, redup } from 'redux-decorator';
class State {
#redup("Todos", "AddTodo", [])
addTodo(state, action) {
return [...state, { id: 2 }];
}
#redup("Todos", "RemoveTodo", [])
removeTodo(state, action) {
console.log("running remove todo");
const copy = [...state];
copy.splice(action.index, 1);
return copy;
}
}
const store = createStore(create_reducer(new State()));
You can also even nest your state:
class Note{
#redup("Notes","AddNote",[])
addNote(state,action){
//Code to add a note
}
}
class State{
aConstant = 1
#redup("Todos","AddTodo",[])
addTodo(state,action){
//Code to add a todo
}
note = new Note();
}
// create store...
//Adds a note
store.dispatch({
type:'AddNote'
})
//Log notes
console.log(store.getState().note.Notes)
Lots of documentation available on NPM. As always, feel free to contribute!

A key decision to be made when designing React/Redux programs is where to put business logic (it has to go somewhere!).
It could go in the React components, in the action creators, in the reducers, or a combination of those. Whether the generic action/reducer combination is sensible depends on where the business logic goes.
If the React components do the majority of the business logic, then the action creators and reducers can be very lightweight, and could be put into a single file as you suggest, without any problems, except making the React components more complex.
The reason that most React/Redux projects seem to have a lot of files for action creators and reducers because some of the business logic is put in there, and so would result in a very bloated file, if the generic method was used.
Personally, I prefer to have very simple reducers and simple components, and have a large number of actions to abstract away complexity like requesting data from a web service into the action creators, but the "right" way depends on the project at hand.
A quick note: As mentioned in https://stackoverflow.com/a/50646935, the object should be returned from setState. This is because some asynchronous processing may need to happen before store.dispatch is called.
An example of reducing boilerplate is below. Here, a generic reducer is used, which reduces code needed, but is only possible the logic is handled elsewhere so that actions are made as simple as possible.
import ActionType from "../actionsEnum.jsx";
const reducer = (state = {
// Initial state ...
}, action) => {
var actionsAllowed = Object.keys(ActionType).map(key => {
return ActionType[key];
});
if (actionsAllowed.includes(action.type) && action.type !== ActionType.NOP) {
return makeNewState(state, action.state);
} else {
return state;
}
}
const makeNewState = (oldState, partialState) => {
var newState = Object.assign({}, oldState);
const values = Object.values(partialState);
Object.keys(partialState).forEach((key, ind) => {
newState[key] = values[ind];
});
return newState;
};
export default reducer;
tldr It is a design decision to be made early on in development because it affects how a large portion of the program is structured.

Performance wise not much. But from a design perspective quite a few. By having multiple reducers you can have separation of concerns - each module only concerned with themselves. By having action creators you add a layer of indirection -allowing you to make changes more easily. In the end it still depends, if you don't need these features a generic solution helps reduce code.

First of all, some terminology:
action: a message that we want to dispatch to all reducers. It can be anything. Usually it's a simple Javascript object like const someAction = {type: 'SOME_ACTION', payload: [1, 2, 3]}
action type: a constant used by the action creators to build an action, and by the reducers to understand which action they have just received. You use them to avoid typing 'SOME_ACTION' both in the action creators and in the reducers. You define an action type like const SOME_ACTION = 'SOME_ACTION' so you can import it in the action creators and in the reducers.
action creator: a function that creates an action and dispatches it to the reducers.
reducer: a function that receives all actions dispatched to the store, and it's responsible for updating the state for that redux store (you might have multiple stores if your application is complex).
Now, to the question.
I think that a generic action creator is not a great idea.
Your application might need to use the following action creators:
fetchData()
fetchUser(id)
fetchCity(lat, lon)
Implementing the logic of dealing with a different number of arguments in a single action creator doesn't sound right to me.
I think it's much better to have many small functions because they have different responsibilities. For instance, fetchUser should not have anything to do with fetchCity.
I start out by creating a module for all of my action types and action creators. If my application grows, I might separate the action creators into different modules (e.g. actions/user.js, actions/cities.js), but I think that having separate module/s for action types is a bit overkill.
As for the reducers, I think that a single reducer is a viable option if you don't have to deal with too many actions.
A reducer receives all the actions dispatched by the action creators. Then, by looking at the action.type, it creates a new state of the store. Since you have to deal with all the incoming actions anyway, I find it nice to have all the logic in one place. This of course starts to be difficult if your application grows (e.g. a switch/case to handle 20 different actions is not very maintainable).
You can start with a single reducer, the move to several reducers and combine them in a root reducer with the combineReducer function.

Testing Complex Asynchronous Redux Actions

So, let's say I have the next action:
export function login({ email, password, redirectTo, doNotRedirect }) {
return ({ dispatch }) => {
const getPromise = async () => {
const basicToken = Base64.encode(`${email}:${password}`);
const authHeaders = { Authorization: `Basic ${basicToken}` };
const { payload, error } = await dispatch(sendAuthentication(authHeaders));
if (error) throw payload;
const { username, token, fromTemporaryPassword } = payload;
const encodedToken = Base64.encode(`${username}:${token}`);
dispatch(persistence.set('authorizationToken', encodedToken));
dispatch(postGlobalId({ username }));
dispatch(setIsLoggedIn(true));
dispatch(setIsFromTemporaryPassword(fromTemporaryPassword));
await dispatch(clientActions.fetchClient);
if (doNotRedirect) return;
if (fromTemporaryPassword)
dispatch(updatePath('/profile/change-password'));
else
dispatch(updatePath(redirectTo || '/dashboard'));
};
return {
type: AUTHENTICATION_LOGIN,
payload: getPromise()
};
};
}
And I want to add tests for it, to add reliability to the code.
So, here are few things:
We send authentication headers and get data as a response
We throw an error if some error is present in the response
We set up all needed tokens, dispatch all needed actions to show that we are logged in now
Fetching client data
Based on params and received data, we redirect to needed route / don't redirect
The question is that it is really too hard to test and we need to stub literally everything, which is bad due to brittle tests, fragility and too much of implementation knowing (not to mention that it is pretty challenging to stub dispatch to work properly).
Therefore, should I test all of these 5 points, or to focus only on the most important stuff, like sending authorization request, throw error and check redirects? I mean, the problem with all flags that they can be changed, so it is not that reliable.
Another solution is just to separate these activities into something like following:
auth
setLoginInfo
handleRedirects
And to pass all needed functions to invoke through dependency injection (here just with params, basically)? With this approach I can spy only invoking of this functions, without going into much details.
I am quite comfortable with unit testing of pure functions and handling different edge-cases for them (without testing too much implementation, just the result), but testing complex functions with side-effects is really hard for me.

If you have very complex actions like that, I think an alternative (better?) approach is to have simple synchronous actions instead (you can even just dispatch payloads directly, and drop action creators if you like, reducing boiler-plate), and handle the asynchronous side using redux-saga: https://github.com/yelouafi/redux-saga
Redux Saga makes it very simple to factor out your business logic code into multiple simple generator functions that can be tested in isolation. They can also be tested without the underlying API methods even being called, due to the 'call' function in that library: http://yelouafi.github.io/redux-saga/docs/api/index.html#callfn-args. Due to the use of generators, your test can 'feed' values to the saga using the standard iterator.next method. Finally, they make it much easier for reducers to have their say, since you can check something from store state (e.g. using a selector) to see what to do next in your saga.
If Redux + Redux Saga had existed before I started on my app (about 100,000 JS(X) LOC so far), I would definitely have used them.

Shortest code to cache Rxjs http request while not complete?

I'm trying to create an observable flow that fulfills the following requirements:
Loads data from storage at subscribe time
If the data has not yet expired, return an observable of the stored value
If the data has expired, return an HTTP request observable that uses the refresh token to get a new value and store it
If this code is reached again before the request has completed, return the same request observable
If this code is reached after the previous request completed or with a different refresh token, start a new request
I'm aware that there are many different answers on how to perform step (3), but as I'm trying to perform these steps together I am looking for guidance on whether the solution I've come up with is the most succinct it can be (which I doubt!).
Here's a sample demonstrating my current approach:
var cachedRequestToken;
var cachedRequest;
function getOrUpdateValue() {
return loadFromStorage()
.flatMap(data => {
// data doesn't exist, shortcut out
if (!data || !data.refreshtoken)
return Rx.Observable.empty();
// data still valid, return the existing value
if (data.expires > new Date().getTime())
return Rx.Observable.return(data.value);
// if the refresh token is different or the previous request is
// complete, start a new request, otherwise return the cached request
if (!cachedRequest || cachedRequestToken !== data.refreshtoken) {
cachedRequestToken = data.refreshtoken;
var pretendHttpBody = {
value: Math.random(),
refreshToken: Math.random(),
expires: new Date().getTime() + (10 * 60 * 1000) // set by server, expires in ten minutes
};
cachedRequest = Rx.Observable.create(ob => {
// this would really be a http request that exchanges
// the one use refreshtoken for new data, then saves it
// to storage for later use before passing on the value
window.setTimeout(() => { // emulate slow response
saveToStorage(pretendHttpBody);
ob.next(pretendHttpBody.value);
ob.completed();
cachedRequest = null; // clear the request now we're complete
}, 2500);
});
}
return cachedRequest;
});
}
function loadFromStorage() {
return Rx.Observable.create(ob => {
var storedData = { // loading from storage goes here
value: 15, // wrapped in observable to delay loading until subscribed
refreshtoken: 63, // other process may have updated this between requests
expires: new Date().getTime() - (60 * 1000) // pretend to have already expired
};
ob.next(storedData);
ob.completed();
})
}
function saveToStorage(data) {
// save goes here
}
// first request
getOrUpdateValue().subscribe(function(v) { console.log('sub1: ' + v); });
// second request, can occur before or after first request finishes
window.setTimeout(
() => getOrUpdateValue().subscribe(function(v) { console.log('sub2: ' + v); }),
1500);

First, have a look at a working jsbin example.
The solution is a tad different then your initial code, and I'd like to explain why. The need to keep returning to your local storage, save it, save flags (cache and token) didn't not fit for me with reactive, functional approach. The heart of the solution I gave is:
var data$ = new Rx.BehaviorSubject(storageMock);
var request$ = new Rx.Subject();
request$.flatMapFirst(loadFromServer).share().startWith(storageMock).subscribe(data$);
data$.subscribe(saveToStorage);
function getOrUpdateValue() {
return data$.take(1)
.filter(data => (data && data.refreshtoken))
.switchMap(data => (data.expires > new Date().getTime()
? data$.take(1)
: (console.log('expired ...'), request$.onNext(true) ,data$.skip(1).take(1))));
}
The key is that data$ holds your latest data and is always up to date, it is easily accessible by doing a data$.take(1). The take(1) is important to make sure your subscription gets a single values and terminates (because you attempt to work in a procedural, as opposed to functional, manner). Without the take(1) your subscription would stay active and you would have multiple handlers out there, that is you'll handle future updates as well in a code that was meant only for the current update.
In addition, I hold a request$ subject which is your way to start fetching new data from the server. The function works like so:
The filter ensures that if your data is empty or has no token, nothing passes through, similar to the return Rx.Observable.empty() you had.
If the data is up to date, it returns data$.take(1) which is a single element sequence you can subscribe to.
If not, it needs a refresh. To do so, it triggers request$.onNext(true) and returns data$.skip(1).take(1). The skip(1) is to avoid the current, out dated value.
For brevity I used (console.log('expired ...'), request$.onNext(true) ,data$.skip(1).take(1))). This might look a bit cryptic. It uses the js comma separated syntax which is common in minifiers/uglifiers. It executes all statements and returns the result of the last statement. If you want a more readable code, you could rewrite it like so:
.switchMap(data => {
if(data.expires > new Date().getTime()){
return data$.take(1);
} else {
console.log('expired ...');
request$.onNext(true);
return data$.skip(1).take(1);
}
});
The last part is the usage of flatMapFirst. This ensures that once a request is in progress, all following requests are dropped. You can see it works in the console printout. The 'load from server' is printed several times, yet the actual sequence is invoked only once and you get a single 'loading from server done' printout. This is a more reactive oriented solution to your original refreshtoken flag checking.
Though I didn't need the saved data, it is saved because you mentioned that you might want to read it on future sessions.
A few tips on rxjs:
Instead of using the setTimeout, which can cause many problems, you can simply do Rx.Observable.timer(time_out_value).subscribe(...).
Creating an observable is cumbersome (you even had to call next(...) and complete()). You have a much cleaner way to do this using Rx.Subject. Note that you have specifications of this class, the BehaviorSubject and ReplaySubject. These classes are worth knowing and can help a lot.
One last note. This was quite a challange :-) I'm not familiar with your server side code and design considerations yet the need to suppress calls felt uncomfortable to me. Unless there is a very good reason related to your backend, my natural approach would be to use flatMap and let the last request 'win', i.e. drop previous un terminated calls and set the value.
The code is rxjs 4 based (so it can run in jsbin), if you're using angular2 (hence rxjs 5), you'll need to adapt it. Have a look at the migration guide.
================ answers to Steve's other questions (in comments below) =======
There is one article I can recommend. It's title says it all :-)
As for the procedural vs. functional approach, I'd add another variable to the service:
let token$ = data$.pluck('refreshtoken');
and then consume it when needed.
My general approach is to first map my data flows and relations and then like a good "keyboard plumber" (like we all are), build the piping. My top level draft for a service would be (skipping the angular2 formalities and provider for brevity):
class UserService {
data$: <as above>;
token$: data$.pluck('refreshtoken');
private request$: <as above>;
refresh(){
request.onNext(true);
}
}
You might need to do some checking so the pluck does not fail.
Then, each component that needs the data or the token can access it directly.
Now lets suppose you have a service that needs to act on a change to the data or the token:
class SomeService {
constructor(private userSvc: UserService){
this.userSvc.token$.subscribe(() => this.doMyUpdates());
}
}
If your need to synthesize data, meaning, use the data/token and some local data:
Rx.Observable.combineLatest(this.userSvc.data$, this.myRelevantData$)
.subscribe(([data, myData] => this.doMyUpdates(data.someField, myData.someField));
Again, the philosophy is that you build the data flow and pipes, wire them up and then all you have to do is trigger stuff.
The 'mini pattern' I've come up with is to pass to a service once my trigger sequence and register to the result. Lets take for example autocomplete:
class ACService {
fetch(text: string): Observable<Array<string>> {
return http.get(text).map(response => response.json().data;
}
}
Then you have to call it every time your text changes and assign the result to your component:
<div class="suggestions" *ngFor="let suggestion; of suggestions | async;">
<div>{{suggestion}}</div>
</div>
and in your component:
onTextChange(text) {
this.suggestions = acSVC.fetch(text);
}
but this could be done like this as well:
class ACService {
createFetcher(textStream: Observable<string>): Observable<Array<string>> {
return textStream.flatMap(text => http.get(text))
.map(response => response.json().data;
}
}
And then in your component:
textStream: Subject<string> = new Subject<string>();
suggestions: Observable<string>;
constructor(private acSVC: ACService){
this.suggestions = acSVC.createFetcher(textStream);
}
onTextChange(text) {
this.textStream.next(text);
}
template code stays the same.
It seems like a small thing here, but once the app grows bigger, and the data flow complicated, this works much better. You have a sequence that holds you data and you can use it around the component wherever you need it, you can even further transform it. For example, lets say you need to know the number of suggestions, in the first method, once you get the result, you need to further query it to get it, thus:
onTextChange(text) {
this.suggestions = acSVC.fetch(text);
this.suggestionsCount = suggestions.pluck('length'); // in a sequence
// or
this.suggestions.subscribe(suggestions => this.suggestionsCount = suggestions.length); // in a numeric variable.
}
Now in the second method, you just define:
constructor(private acSVC: ACService){
this.suggestions = acSVC.createFetcher(textStream);
this.suggestionsCount = this.suggestions.pluck('length');
}
Hope this helps :-)
While writing, I tried to reflect about the path I took to getting to use reactive like this. Needless to say that on going experimentation, numerous jsbins and strange failures are big part of it. Another thing that I think helped shape my approach (though I'm not currently using it) is learning redux and reading/trying a bit of ngrx (angular's redux port). The philosophy and the approach does not let you even think procedural so you have to tune in to functional, data, relations and flows based mindset.

Where should I put synchronous side effects linked to actions in redux?

(Note: My question was not clearly written, and I was thinking about some things wrong. The current version of the question is just an attempt to write something that could make the accepted answer useful to as many people as possible.)
I want to have an action that adds an item to a store and registers it with an external dependency.
I could use the thunk middleware and write
export function addItem(item) {
return dispatch => {
dispatch(_addItemWithoutRegisteringIt(item));
externalDependency.register(item);
};
}
But the subscribers would be notified before the item was registered, and they might depend on it being registered.
I could reverse the order and write
export function addItem(item) {
return dispatch => {
externalDependency.register(item);
dispatch(_addItemWithoutRegisteringIt(item));
};
}
But I track the item in the external dependency by a unique id that it is natural to only assign in the reducer.
I could register the item in the reducer, but I am given to understand that it is very bad form to do side effects in a reducer and might lead to problems down the line.
So what is the best approach?
(My conclusion is: there are a number of approaches that would work, but probably the best one for my use case is to store a handle into the external dependency in Redux rather than a handle into Redux in the external dependency.)

If you use Redux Thunk middleware, you can encapsulate it in an action creator:
function addItem(id) {
return { type: 'ADD_ITEM', id };
}
function showNotification(text) {
return { type: 'SHOW_NOTIFICATION', text };
}
export function addItemWithNotification(id) {
return dispatch => {
dispatch(addItem(id));
doSomeSideEffect();
dispatch(showNotification('Item was added.');
};
}
Elaborating, based on the comments to this answer:
Then maybe this is the wrong pattern for my case. I don't want subscribers invoked between dispatch(addItem(id)) and doSomeSideEffect().
In 95% cases you shouldn't worry about whether the subscribers were invoked. Bindings like React Redux won't re-render if the data hasn't changed.
Would putting doSomeSideEffect() in the reducer be an acceptable approach or does it have hidden pitfalls?
No, putting side effects into the reducer is never acceptable. This goes against the central premise of Redux and breaks pretty much any tool in its ecosystem: Redux DevTools, Redux Undo, any record/replay solution, tests, etc. Never do this.
If you really need to perform a side effect together with an action, and you also really care about subscribers only being notified once, just dispatch one action and use [Redux Thunk] to “attach” a side effect to it:
function addItem(id, item) {
return { type: 'ADD_ITEM', id, item };
}
export function addItemWithSomeSideEffect(id) {
return dispatch => {
let item = doSomeSideEffect(); // note: you can use return value
dispatch(addItem(id, item));
};
}
In this case you'd need to handle ADD_ITEM from different reducers. There is no need to dispatch two actions without notifying the subscribers twice.
Here is the one point I still definitely don't understand. Dan suggested that the thunk middleware couldn't defer subscriber notification because that would break a common use case with async requests. I still don't understand this this.
Consider this:
export function doSomethinAsync() {
return dispatch => {
dispatch({ type: 'A' });
dispatch({ type: 'B' });
setTimeout(() => {
dispatch({ type: 'C' });
dispatch({ type: 'D' });
}, 1000);
};
}
When would you want the subscriptions to be notified? Definitely, if we notify the subscribers only when the thunk exits, we won't notify them at all for C and D.
Either way, this is impossible with the current middleware architecture. Middleware isn't meant to prevent subscribers from firing.
However what you described can be accomplished with a store enhancer like redux-batched-subscribe. It is unrelated to Redux Thunk, but it causes any group of actions dispatched synchronously to be debounced. This way you'd get one notification for A and B, and another one notification for C and D. That said writing code relying on this behavior would be fragile in my opinion.

I'm still in the process of learning Redux; however my gut instinct says that this is could be a potential candiate for some custom middleware?