I created this plunker.
The gist is this code:
var myObj = function (){
return {
getThis: function (callback) {
return callback("this");
}
}
}
var myDidSome = function () {
return {
DoSome: function (text) {
return text + " is cool";
}
};
};
var subObj = function (objG, didSome) {
var txt = objG.getThis(function (text) {
return didSome.DoSome(text);
});
console.log(txt);
}
new subObj(new myObj, new myDidSome);
So to deconstruct this code, view myObj and myDidSome as services (angular ones). I'm most interested in the myObj.getThis, this is supposed to mimick some $http response. So when its finished call my callback function passing in the parameters obtained from $http.
The thing I keep thinking about though is $http is async. Will my txt property in the subOj (think of this as a controller) class be populated correctly when the data gets back, I assume it will be undefined till then?
In my scenario this is completely synchronous, which obviously works.
So, I would like to know whether async will be an issue here? And whether doing this practice of propagating the return of an object is a good one, or doesn't matter, or is bad and I shouldn't do it?
You cannot use return like that with an asynchronous method. You have to do everything within the callback. In your example, the result of didSome.DoSome(text) will never be assigned to txt, not even after the callback has run.
For a simple case like this where you're logging the response, doing it in the callback is an acceptable approach. If you need to do something more complex, however, you should look into promises via $q. These provide a much nicer way to rewrite a long chain of callbacks into a more linear flow.
Related
I have been looking for assistance with setting up an IndexedDB for web storage, and I have run into a problem that I cannot find a good answer to. After I have successfully setup/opened the database I am having trouble passing the variable that contains the database information to access it later. Here is my code, I have been following a guide from MDN :
const DB_NAME = 'database-name';
const DB_VERSION = 2;
const DB_STORE_NAME = 'users';
var db;
function openDb() {
console.log('openDb ...')
var request = indexedDB.open(DB_NAME, DB_VERSION);
request.onsuccess = function(event) {
db = request.result;
console.log("openDb DONE");
};
request.onerror = function(event) {
console.log(request.errorCode);
};
request.onupgradeneeded = function(event) {
console.log('openDb.onupgradeneeded');
var store = event.currentTarget.result.createObjectStore(DB_STORE_NAME, { keyPath: 'id', autoIncrement: true });
store.createIndex('age', 'age', { unique: false });
};
}
function getObjectStore(store_name, mode) {
var tx = db.transaction(store_name, mode);
return tx.objectStore(store_name);
}
When getObjectStore is called the variable db is undefined. My knowledge of javascript is very limited and some concepts I don't get. The guide doesn't not show anything special being done and their demo works as is. Some other guides have mentioned implementing a callback, but they don't show how its done nor do I understand the concept of callbacks. Any help is appreciated. Thanks.
Unfortunately you need to learn about a relatively complicated concept usually referred to as asynchronous JavaScript before proceeding to use indexedDB. There are already several thousand questions related to AJAX on stackoverflow. I am trying to think of the most polite way to say this, but basically, the answer you are looking for is already provided by these other questions, and by many other websites. Nevertheless, here are some quick tips.
First, your approach is never going to work. You cannot skip over learning about async.
Second, do not use the setTimeout trick to get it to work. That is horrible advice.
Third, at a general level, a callback is simply a word used to describe a function when the function is used in a particular way. Specifically, a callback refers to a function that is passed as an argument to another function, where the other function then maybe calls the function at some later point in time. More specifically, a callback is generally a function that is called at the end of the function it is passed to, when the function has completed.
For example:
function a(b) { alert(b); }
function c(d) { d('hi'); }
c(a);
That might look a bit confusing at first but it is the simplest thing I can describe off the top of my head. In the example, the final line calls function c and passes in function a. The effect of the code is that you see 'hi' as a browser alert. In this example, the function a is passed as a parameter/argument to function c. Function c uses the name d for its one and only argument. c calls d with the string 'hi'. When describing this example, we would say that argument d represents a callback function passed to function c. We could also say that function a is the particular callback function used by function c. So that is basically it. When you pass a function in as an argument and the other function calls the passed in argument, you are using a callback.
Then things gets way more complicated, because you have to learn about how to read and write asynchronous code. Properly introducing it would take several pages. Here is an extreme crash course.
You have traditionally been writing synchronous code, even if you did not call it that. Synchronous code runs exactly when you expect it to, in the order that you write your statements. Here is a brief example of typical sync code:
function sum(a, b) { return a + b; }
alert(sum(1, 2));
Simple stuff. The next example is code that uses a callback, but is still synchronous.
function doOperation(op, num1, num2) { return op(num1, num2); }
function sumOperation(num1, num2) { return num1 + num2; }
var result = doOperation(sumOperation, 1, 2);
alert(result);
Here we passed the sumOperation function into the doOperation function. sumOperation is the callback function. It is the first argument with the name 'op'. Still pretty simple stuff. Now consider the next example. The point of the next example is to show how we pass control to the function to do something. Kind of like how goto/labels work.
function doOperation(op, num1, num2) {
var result = op(num1, num2);
alert(result);
return undefined;
}
function sumOperation(num1, num2) { return num1 + num2; }
doOperation(sumOperation, 1, 2);
Notice how doOperation no longer returns a value. It has the logic within its function body. So once we call doOp, the browser starts running the code inside doOperation. So we switched from the outer context into the body of the function. Also, because doOperation doesn't return anything, we cannot do anything with its return value. The logic is locked inside the body of the doOperation function. The code still works about the same, its just that now we are not returning anything from doOperation, and now the logic is inside doOperation instead of outside in the main/global context.
Now an example that uses setTimeout. This is completely unrelated to the suggestion of using setTimeout.
function doOperation(op, num1, num2) {
setTimeout(function runLater() {
var result = op(num1, num2);
alert(result);
return undefined;
}, 1000);
return undefined;
}
function sumOperation(num1, num2) { return num1 + num2; }
doOperation(sumOperation, 1, 2);
The point here is to understand that we use a callback (named runLater in this example), and that the code inside the callback does not run immediately. Therefore, we can no longer say it runs synchronously. We instead refer to the statements constituting the body of the callback function as asynchronous. So now an alert appears after 1 second. Notice how we cannot return anything from runLater. Also notice how we cannot return anything from doOperation. There is nothing to return. There is no way to get the value in the 'result' variable out of the scope of runLater. It is locked in there.
Let's try almost the same thing, but try to have runLater set a variable. Also, I am going to omit 'return undefined' because that is what every function without a explicit return statement returns.
var aGlobalResult = null;
function doOperation(op, num1, num2) {
setTimeout(function runLater() {
aGlobalResult = op(num1, num2);
}, 1000);
}
function sumOperation(num1, num2) { return num1 + num2; }
doOperation(sumOperation, 1, 2);
alert(aGlobalResult);
Hopefully you are catching on to the problem. First off, runLater does not return anything, so doOperation does not return anything, so we could not even try to do something like aGlobalResult = doOperation(...);, because that would not make any sense. Second, the result here is that you will see an alert 'undefined' because the alert statement executes prior to the statement that assigns a value to aGlobalResult. This is even though you wrote the assignment statement higher up (earlier) in the code, and the alert is later. This is the brick wall some newer developers run into right here. This is indeed confusing for some. aGlobalResult is undefined here because setTimeout does not set it until later. Even if we passed in 0 milliseconds to setTimeout, it is still 'later', meaning that the assignment happens at a later point in time, after the alert. The alert message is always going to be undefined. There is absolutely nothing you can do to avoid the way this works. Nothing. Period. Stop trying. Learn it, or give up entirely.
So, how does one typically write code that behaves or involves asynchronous stuff? By using callbacks. Which again means that you can no longer use return statements to assign values to outer scope variables. You instead want to write functions and pass around control to various functions. In other words, instead of:
function a() {}
function b() {}
function c() {}
a(); b(); c();
you write code like this:
function a(callback) {
var asdf = 1+2; // do some stuff in a
alert('a finished');
// a has now completed, call its callback function, appropriately named callback
callback();
}
function b(callback) {
var asdfasdfasdf = 3 + 4;
alert('b finished');
// call the callback
callback();
}
a( function(){ b(function() { alert('both a and b finished'); }); });
This is more formally known as continuation passing style, or CPS.
So, that is an example of the very basics of writing callback functions and basic asynchronous code. Now you can start to use indexedDB. The first thing you will notice is that the function indexedDB.open is documented as asynchronous. So, how can we use it? Like this:
var someGlobalVariable = null;
var openRequest = indexedDB.open(...);
openRequest.onsuccess = function openRequestOnSuccessCallbackFunction(event) {
// Get a reference to the database variable. If this function is ever called at some later
// point in time, the database variable is now defined as the 'result' property of the
// open request. There are multiple ways to access the result property. Any of the following
// lines works 100% the same way. Use whatever you prefer. It does not matter.
var databaseConnection = openRequest.result;
var databaseConnection = this.result;
var databaseConnection = event.target.result;
var databaseConnection = event.currentTarget.result;
// Now that we have a valid and defined databaseConnection variable, do something with it
// e.g.:
var transaction = databaseConnection.transaction(...);
var objectStore = transaction.objectStore(...);
// etc.
// DO NOT DO THE FOLLOWING, it does not work. Why? Review the early descriptions. First off
// this onsuccess callback function does not return anything. Second off this function gets
// called at some *later* point in time, who knows when. It could be a nanosecond later.
someGlobalVariable = databaseConnection;
};
Hopefully that sets you on the path.
Edit: I thought I would add a bit more of an intro. A related concept you need to learn that I did not explain clearly enough regarding control is the difference between imperative and declarative programming.
Imperative programming involves executing a series of statements in the order you wrote. You are the caller and are in control. Imperative code looks like this (fictional code):
var dbConn = dbFactory.newConnection('mydb');
var tx = dbConn.newTransaction();
var resultCode = tx.put(myObject);
if(resultCode == dbResultConstants.ERROR_PUT_KEY_EXISTS) {
alert('uhoh');
}
Declarative programming is subtly different. With a declarative approach, you write functions, and then you register (aka hook or bind) the functions to the JavaScript engine, and then at some point later, when it is appropriate, the engine runs your code. The engine is the caller and is in control, not you. Declarative programming involves callbacks and looks like this (fictional code):
dbFactory.newConnection(function onConnect(dbConn) {
dbConn.newTransaction(function onNewTransaction(tx) {
tx.put(myObject, function onPut(resultCode) {
if(resultCode == dbResultConstants.ERROR_PUT_KEY_EXISTS) {
alert('uhoh');
}
});
});
});
In this example, the only thing you called was the fictional dbFactory.newConnection function. You passed in a callback function. You did not call the callback function yourself. The engine calls the callback function. You cannot call the callback function yourself. This is kind of the whole idea behind why JavaScript engines can allow you to write asynchronous code. Because you don't get to control the order of execution of statements/functions. The engine gets to control it. All you get to do is write your functions, register them, and then start a chain of callbacks (the sole imperative line, the starting statement).
So this is why a function like getObjectStore in your question will not work. You are trying to call the function yourself, but that is backwards. You can only write a function and register it (somehow hook it up as a callback somewhere) and then the engine, not you, calls it at some later point in time.
Hopefully this is not more confusing, but you could actually write your function getObjectStore if you really wanted to by passing in the database variable to the function as its first argument. This leads to the logical next question, how to get a valid database variable to pass into the function. You cannot get one in a global context (reliably). Because the connection variable is only valid within the context of the onOpen callback function. So you would have to make your call to this function from within the onOpen function. Something like:
function getObjectStore(db, name, mode) {
var tx = db.transaction(name, mode);
var store = tx.objectStore(name);
return store;
}
var openRequest = indexedDB.open(...);
openRequest.onsuccess = function onOpen(event) {
// get the connection variable. it is defined within this (onOpen) function and open.
var db = this.result;
// call our simple imperative helper function to get the users store. only call it from
// within this onOpen function because that is the only place we can get the 'db' variable.
var usersStore = getObjectStore(db, 'users', 'readwrite');
// do something here with usersStore, inside this function only.
};
I'm having an issue that is burning my head. Basically this is the scenario:
I have a callController() function, which just simple use the jQuery.load() method to load a controller, after loaded, I can play with the returning parameters.
Now... in one of my controllers there is a validation rule I need to check in order to to allow the user to execute certain functionality, however, I create a function like this:
function myValRule() {
var val = callController('blablabla'),
response = "";
val.done(function(data) {
//my business logic
response = something;
}
return response;
}
As you imagine, response, even if It has a value, it returns undefined, so I set a timeout and console.log() it and now it has a value, but I cannot make to return this value even if I put the return into the setTimeout(). So basically when the method call this function to validate it finds it empty.
Can you point me on some direction to solve this issue?
Remember this is asynchronous! The return response; is long gone by the time .done() is actually called.
if you need the value returned in this fashion, look at either supplying a callback function in your myValRule function (that would be called from within .done()) or maybe look at using the $.Deferred api so you can mimic what callController(...) is doing.
Here's an example of both scenarios (with example call to myValRule):
Callback argument
function myValRule(callback){
var val = callController('blablabla'),
val.done(function(data){
var response = /* something */;
callback(response);
});
}
myValRule(function(response){
// here you have response
});
Using $.Deferred
(I assume it's jQuery since there's a .done call)
function myValRule(){
var df = $.Deferred(),
val = callController('blablabla');
val.done(function(data){
var response = /*something */;
df.resolve(response);
}).fail(df.reject);
return df.promise();
}
myValRule().done(function(response){
// here you have response
});
I'm still trying to get my head around all this and I'm obviously missing some fundamental concepts.
In my code I have a scenario whereby I want to call several functions in a row and when they've all completed, wrap it up with a finishing routine where subtotals are calculated and a pie chart is updated.
Although the functions I call themselves aren't asynchronous, they do contain ajax calls which are, hence why I want to wait for all of them to finish before I calculate totals and update the chart.
So I tried doing:
var fnArray= [];
fnArray.push(genericCalc("use_", true, false, false));
fnArray.push(doArticleImpacts(false));
fnArray.push(genericProjectCalc("tpD2C_", false, false));
fnArray.push(genericCalc("eol_", true, false, false));
fnArray.push(calcPackaging(false));
var calcPromise = Q.all(fnArray);
return calcPromise
.then(calcsDone)
.fail(calcsFailed);
function calcsDone() {
calcTotals();
setChart(selectedRow());
}
function calcsFailed() {
logger.logError("Failure in calculations", "", null, true);
}
...but using the above code and using the script debugger with a stop on the "return calcPromise" line, the fnArray is set to "0:undefined, 1:undefined, 2:Object, 3:undefined, 4:Promise" even before the promise is activated.
I understand that this is obviously something to do with my functions, but I don't really understand what I need to do differently. The functions all vary slightly, but are fundamentally something like:
var genericCalc = function (calcPrefix) {
var res_Array = ko.observable(); //holds returned results
prjArticleArray.forEach(function (thisArticle) {
var calcPromise = calcEOL(res_Array, thisArticle); //another function containing async ajax call
return calcPromise
.then(calcsDone)
.fail(calcsFailed);
function calcsDone() {
//do calculation subtotals here and set a knockout observable value
}
function calcsFailed() {
logger.logError("Failure in " + calcPrefix + "calculation", "", null, true);
}
});
};
What is it that makes some of the functions "undefined", some "object" and some "promise" in my array that I want to use for Q.all? Do I have to have "Q.resolve" in the "calcsDone" part of the functions I'm calling?
I've seen other questions/answers on stackoverflow along similar lines, but they seem to always be calls directly to async calls and my fist level functions I'm stacking up in the promise aren't... should I not be using this structure for non-async calls or just add "setTimeout" to my function calls to make them async?
I understand that this is obviously something to do with my functions, but I don't really understand what I need to do differently
You need to return promises from them. Your current genericCalc function does not even contain a return statement (only the forEach callback function inside it does), so it does return undefined.
If you need to wait for the result of each article, use Q.all like you have already done with fnArray (which is a promiseArray actually). In genericCalc it should look like this:
var genericCalc = function (calcPrefix) {
return Q.all(prjArticleArray.map(function (thisArticle) {
var calcPromise = calcEOL(res_Array, thisArticle); //another function containing async ajax call
var result = calcPromise
.then(calcsDone)
.fail(calcsFailed);
return result;
…
});
};
var res_Array = ko.observable(); //holds returned results
…
//do calculation subtotals here and set a knockout observable value
That is a bad idea. You should not use global variables that are set somewhen and use promises just for propagating changes, but the promises should represent these values. This leads to a better, functional programming style.
So instead of setting a global variable, do return the result from your calcsDone function, and it will resolve the result promise with that value.
I've been struggling for awhile with callbacks in general. I'm trying to go back to the basics to try to understand conceptually and this is what I understand so far (yeah, basic).
function RandHash (callback) {
ds_hash = Math.floor((Math.random()*10000)+1);
callback();
}
function CompiledHash (){
console.log(ds_hash);
}
var ds_hash;
RandHash(Compiled Hash);
Will yield the random number.
However, I'm lost as to how to get the "ds_hash" variable returned from the callback.
Seems this would work:
var ds_hash;
ds_hash = RandHash(Compiled Hash);
It doesn't. If I try to return the value something like:
function CompiledHash (){
return ds_hash;
}
It doesn't do anything.
Please help me with this. Seems I spend 90% of my time with node in callback debugging. I've built some decent applications but everything has been handled through the async library because of this mental block I have.
Thanks.
The first error you've made is that RandHash hasn't returned anything. There's no return statement present in the function, so var anything = RandHash(alsoAnything) will always result in anything being undefined.
Functions are first class variables
Functions can be used just as variables can, and passed round as arguments to functions. This is a powerful feature of javascript and something that you'll need to grok to use callbacks. Think of a function as a defined action, and you're just passing that action around.
Also, callbacks should be used to deal with the completion of a process. So the idea is, you know what is meant to happen after process A, and what it is meant to generate, so you can give process A a callback which will be called once A has terminated. Now the scenario here isn't appropriate for a callback situation. It would be much easier for you to do...
function RandHash () {
return Math.floor((Math.random()*10000)+1);
}
And then get your hash by just calling this function, like so...
var hash = RandHash();
You also want to be aware of javascripts variable scoping, as you're missing the var keyword when referencing ds_hash in the RandHash function which means the assignment defaults to global scope. This is probably what is responsible for confusing you, as your assignment of ds_hash in RandHash will be global and therefore available in CompiledHash, meaning some functions will still be able to access the ds_hash value despite this not being the correct, or proper way to do this.
Assuming you will eventually require asynchronous processing
function randHash (callback) {
var hash = /* do hash calculation */
callback(hash);
}
function compileHash (hash) {
/* do something using the hash variable passed to this function */
}
randHash(compileHash); // pass the compileHash function as the callback
You should pass your variables as arguments to the callback. That will deal with your scoping issues hopefully.
Also, small note, functions in javascript should typically be lowercase if they aren't going to be used with the new statement (ie, a javascript class).
To truly understand asynchronous behaviour, you should try it with something that is actually asynchronous, and passing values around, not just a global variable, as that will never work when you move on to asynchronous functions :
function RandHash (callback) {
setTimeout(function() {
var ds_hash = Math.floor((Math.random()*10000)+1);
callback(ds_hash); // execute callback when async operation is done
}, 300);
}
function CompiledHash(hash){ // pass hash
console.log(hash);
}
RandHash(function(returned_hash) { // this is the callback function
CompiledHash(returned_hash); // pass along the returned data
});
The callback function is executed once the asynchronous function, in this case setTimeout, has completed, and passes the data back as an argument.
The callback argument is just the function that is passes as an argument.
// regular function
function doStuff(argument) {
// where argument can be anything, also a function
}
// you can pass a string
doStuff('string');
// or a function
doStuff(function(argument_returned) {
});
// and passing a function you can execute that function with arguments in the original function
function doStuff(argument) {
var argument_returned = 'string';
argument(argument_returned); // executes the passsed function
});
Try this:
function RandHash (callback) {
ds_hash = Math.floor((Math.random()*10000)+1);
callback(ds_hash);
}
function CompiledHash(ds_hash){
console.log(ds_hash);
}
var ds_hash;
RandHash(CompiledHash);
For callbacks and functions, it's a bit difficult to understand variable scopes, and what you are doing is not really recommended. You can pass parameters into callbacks, and should get passed, but I would suggest building it like this:
function RandHash (callback) {
var ds_hash = Math.floor((Math.random()*10000)+1);
callback(ds_hash);
}
function CompiledHash(ds_hash){
console.log(ds_hash);
}
RandHash(CompiledHash);
A couple notes:
If you want to propagate the result of the callback then you need to return its output in RandHash
You don't need ds_hash to be a global variable if you're planning on returning it, anyway
The line RandHash(Compiled Hash); is a syntax error (notice the space in the variable name)
Try this:
function RandHash(callback) {
var ds_hash = Math.floor((Math.random() * 10000) + 1);
return callback(ds_hash);
}
function CompiledHash(ds_hash) {
console.log(ds_hash);
return ds_hash;
}
RandHash(CompiledHash);
Notice that there are no global variables. The callback returns a value and the function that executes the callback passes it along.
On a style note, you should only capitalize the names of functions that you intend to use as a constructor.
This will do
function RandHash (callback) {
var ds_hash = Math.floor((Math.random()*10000)+1);
callback(ds_hash);
}
var CompiledHash = function (ds_hash){
console.log(ds_hash);
}
RandHash(CompiledHash);
Kept on seeing this pattern in code, but couldn't find any reference to it in google or SO, strange. Can someone point me to reference for this.async() function?
var done = this.async();
// ...
$.get(path, function(contents) { // or some other function with callback
// ...
done(JST[path] = tmpl);
})
var done = this.async() and done(blah) is a clever trick to return a value fetched from asynchronous call (e.g. $.get) within a synchronous function.
Let's see an example:
var getText = function() {
return "hello";
};
var text = getText();
It's a pretty straightforward function call so no puzzle here. However, what if you need to fetch the text asynchronously in getText() function?
var getText = function() {
return $.get('<some-url>', function(text) {
return text;
}); // ??????
};
call to getText() doesn't return the text you want to get. It returns jquery's promise object.
So how do we make getText() return the text it gets from $.get() call?
var getText = function() {
var done = this.async();
$.get('<some-url>', function(text) {
done(text);
});
};
var text = getText(); // you get the expected text
Magic, right?
I don't know the inner-working of this.async() call yet. I don't know if there is a library provides that function, but you can see that Backbone.LayoutManager uses this trick https://github.com/tbranyen/backbone.layoutmanager/blob/master/backbone.layoutmanager.js (search for this.async).
Also, Tim Branyen (the author of backbone layoutmanager) briefly talks about it in his video tutorial (http://vimeo.com/32765088 around 14:00 - 15:00). In the video, Tim says Ben Alman came up with that trick. Take a look at this as well https://github.com/cowboy/javascript-sync-async-foreach
I think it's a pretty neat trick to mix async and sync functions.
Cheers,
var done = this.async() is a pattern used in Grunt to help perform asynchronous functions within a Task.
You need to invoke done() or done(returnValues) to tell Grunt the task is complete (after your chain of asynchronous tasks).
Read more about it:
https://gruntjs.com/inside-tasks#inside-all-tasks
It is a way to work around the problem of this escaping inside callback. Without this extra reference the code would look like this:
$.get(path, function(contents) { // or some other function with callback
//Wrong! `this` might no longer point to your object
this.done(JST[path] = tmpl);
})
Unfortunately! this inside response callback is not the same as this outside of it. In fact it can be anything, depending on what $.get (calling the callback using) decides it to be. Most of the people use extra reference named that for the same purpose:
var that = this;
// ...
$.get(path, function(contents) { // or some other function with callback
// ...
that.async(JST[path] = tmpl);
})
This pattern also seems reasonable and readable.
Oh, and if you are curious about this syntax:
done(JST[path] = tmpl)
This is an assignment used as an expression. The value of assignment is the right-hand side, so this code is equivalent to:
JST[path] = tmpl;
done(tmpl);