We all know that global variables are anything but best practice. But there are several instances when it is difficult to code without them. What techniques do you use to avoid the use of global variables?
For example, given the following scenario, how would you not use a global variable?
JavaScript code:
var uploadCount = 0;
window.onload = function() {
var frm = document.forms[0];
frm.target = "postMe";
frm.onsubmit = function() {
startUpload();
return false;
}
}
function startUpload() {
var fil = document.getElementById("FileUpload" + uploadCount);
if (!fil || fil.value.length == 0) {
alert("Finished!");
document.forms[0].reset();
return;
}
disableAllFileInputs();
fil.disabled = false;
alert("Uploading file " + uploadCount);
document.forms[0].submit();
}
Relevant markup:
<iframe src="test.htm" name="postHere" id="postHere"
onload="uploadCount++; if(uploadCount > 1) startUpload();"></iframe>
<!-- MUST use inline JavaScript here for onload event
to fire after each form submission. -->
This code comes from a web form with multiple <input type="file">. It uploads the files one at a time to prevent huge requests. It does this by POSTing to the iframe, waiting for the response which fires the iframe onload, and then triggers the next submission.
You don't have to answer this example specifically, I am just providing it for reference to a situation in which I am unable to think of a way to avoid global variables.
The easiest way is to wrap your code in a closure and manually expose only those variables you need globally to the global scope:
(function() {
// Your code here
// Expose to global
window['varName'] = varName;
})();
To address Crescent Fresh's comment: in order to remove global variables from the scenario entirely, the developer would need to change a number of things assumed in the question. It would look a lot more like this:
Javascript:
(function() {
var addEvent = function(element, type, method) {
if('addEventListener' in element) {
element.addEventListener(type, method, false);
} else if('attachEvent' in element) {
element.attachEvent('on' + type, method);
// If addEventListener and attachEvent are both unavailable,
// use inline events. This should never happen.
} else if('on' + type in element) {
// If a previous inline event exists, preserve it. This isn't
// tested, it may eat your baby
var oldMethod = element['on' + type],
newMethod = function(e) {
oldMethod(e);
newMethod(e);
};
} else {
element['on' + type] = method;
}
},
uploadCount = 0,
startUpload = function() {
var fil = document.getElementById("FileUpload" + uploadCount);
if(!fil || fil.value.length == 0) {
alert("Finished!");
document.forms[0].reset();
return;
}
disableAllFileInputs();
fil.disabled = false;
alert("Uploading file " + uploadCount);
document.forms[0].submit();
};
addEvent(window, 'load', function() {
var frm = document.forms[0];
frm.target = "postMe";
addEvent(frm, 'submit', function() {
startUpload();
return false;
});
});
var iframe = document.getElementById('postHere');
addEvent(iframe, 'load', function() {
uploadCount++;
if(uploadCount > 1) {
startUpload();
}
});
})();
HTML:
<iframe src="test.htm" name="postHere" id="postHere"></iframe>
You don't need an inline event handler on the <iframe>, it will still fire on each load with this code.
Regarding the load event
Here is a test case demonstrating that you don't need an inline onload event. This depends on referencing a file (/emptypage.php) on the same server, otherwise you should be able to just paste this into a page and run it.
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
<title>untitled</title>
</head>
<body>
<script type="text/javascript" charset="utf-8">
(function() {
var addEvent = function(element, type, method) {
if('addEventListener' in element) {
element.addEventListener(type, method, false);
} else if('attachEvent' in element) {
element.attachEvent('on' + type, method);
// If addEventListener and attachEvent are both unavailable,
// use inline events. This should never happen.
} else if('on' + type in element) {
// If a previous inline event exists, preserve it. This isn't
// tested, it may eat your baby
var oldMethod = element['on' + type],
newMethod = function(e) {
oldMethod(e);
newMethod(e);
};
} else {
element['on' + type] = method;
}
};
// Work around IE 6/7 bug where form submission targets
// a new window instead of the iframe. SO suggestion here:
// http://stackoverflow.com/q/875650
var iframe;
try {
iframe = document.createElement('<iframe name="postHere">');
} catch (e) {
iframe = document.createElement('iframe');
iframe.name = 'postHere';
}
iframe.name = 'postHere';
iframe.id = 'postHere';
iframe.src = '/emptypage.php';
addEvent(iframe, 'load', function() {
alert('iframe load');
});
document.body.appendChild(iframe);
var form = document.createElement('form');
form.target = 'postHere';
form.action = '/emptypage.php';
var submit = document.createElement('input');
submit.type = 'submit';
submit.value = 'Submit';
form.appendChild(submit);
document.body.appendChild(form);
})();
</script>
</body>
</html>
The alert fires every time I click the submit button in Safari, Firefox, IE 6, 7 and 8.
I suggest the module pattern.
YAHOO.myProject.myModule = function () {
//"private" variables:
var myPrivateVar = "I can be accessed only from within YAHOO.myProject.myModule.";
//"private" method:
var myPrivateMethod = function () {
YAHOO.log("I can be accessed only from within YAHOO.myProject.myModule");
}
return {
myPublicProperty: "I'm accessible as YAHOO.myProject.myModule.myPublicProperty."
myPublicMethod: function () {
YAHOO.log("I'm accessible as YAHOO.myProject.myModule.myPublicMethod.");
//Within myProject, I can access "private" vars and methods:
YAHOO.log(myPrivateVar);
YAHOO.log(myPrivateMethod());
//The native scope of myPublicMethod is myProject; we can
//access public members using "this":
YAHOO.log(this.myPublicProperty);
}
};
}(); // the parens here cause the anonymous function to execute and return
First off, it is impossible to avoid global JavaScript, something will always be dangling the global scope. Even if you create a namespace, which is still a good idea, that namespace will be global.
There are many approaches, however, to not abuse the global scope. Two of the simplest are to either use closure, or since you only have one variable you need to keep track of, just set it as a property of the function itself (which can then be treated as a static variable).
Closure
var startUpload = (function() {
var uploadCount = 1; // <----
return function() {
var fil = document.getElementById("FileUpload" + uploadCount++); // <----
if(!fil || fil.value.length == 0) {
alert("Finished!");
document.forms[0].reset();
uploadCount = 1; // <----
return;
}
disableAllFileInputs();
fil.disabled = false;
alert("Uploading file " + uploadCount);
document.forms[0].submit();
};
})();
* Note that incrementing of uploadCount is happening internally here
Function Property
var startUpload = function() {
startUpload.uploadCount = startUpload.count || 1; // <----
var fil = document.getElementById("FileUpload" + startUpload.count++);
if(!fil || fil.value.length == 0) {
alert("Finished!");
document.forms[0].reset();
startUpload.count = 1; // <----
return;
}
disableAllFileInputs();
fil.disabled = false;
alert("Uploading file " + startUpload.count);
document.forms[0].submit();
};
I'm not sure why uploadCount++; if(uploadCount > 1) ... is necessary, as it looks like the condition will always be true. But if you do need global access to the variable, then the function property method I described above will allow you to do so without the variable actually being global.
<iframe src="test.htm" name="postHere" id="postHere"
onload="startUpload.count++; if (startUpload.count > 1) startUpload();"></iframe>
However, if that's the case, then you should probably use an object literal or instantiated object and go about this in the normal OO way (where you can use the module pattern if it strikes your fancy).
Sometimes it makes sense to have global variables in JavaScript. But don't leave them hanging directly off window like that.
Instead, create a single "namespace" object to contain your globals. For bonus points, put everything in there, including your methods.
window.onload = function() {
var frm = document.forms[0];
frm.target = "postMe";
frm.onsubmit = function() {
frm.onsubmit = null;
var uploader = new LazyFileUploader();
uploader.startUpload();
return false;
}
}
function LazyFileUploader() {
var uploadCount = 0;
var total = 10;
var prefix = "FileUpload";
var upload = function() {
var fil = document.getElementById(prefix + uploadCount);
if(!fil || fil.value.length == 0) {
alert("Finished!");
document.forms[0].reset();
return;
}
disableAllFileInputs();
fil.disabled = false;
alert("Uploading file " + uploadCount);
document.forms[0].submit();
uploadCount++;
if (uploadCount < total) {
setTimeout(function() {
upload();
}, 100);
}
}
this.startUpload = function() {
setTimeout(function() {
upload();
}, 100);
}
}
Some things are going to be in the global namespace -- namely, whatever function you're calling from your inline JavaScript code.
In general, the solution is to wrap everything in a closure:
(function() {
var uploadCount = 0;
function startupload() { ... }
document.getElementById('postHere').onload = function() {
uploadCount ++;
if (uploadCount > 1) startUpload();
};
})();
and avoid the inline handler.
Other way to do this is to create an object and then add methods to it.
var object = {
a = 21,
b = 51
};
object.displayA = function() {
console.log(object.a);
};
object.displayB = function() {
console.log(object.b);
};
In this way, only object 'obj' is exposed and methods attached to it. It is equivalent to adding it in namespace.
Using closures might be OK for small to medium projects. However, for big projects, you might want to split your code into modules and save them in different files.
Therefore I wrote jQuery Secret plugin to solve the problem.
In your case with this plugin the code would look something like the following.
JavaScript:
// Initialize uploadCount.
$.secret( 'in', 'uploadCount', 0 ).
// Store function disableAllFileInputs.
secret( 'in', 'disableAllFileInputs', function(){
// Code for 'disable all file inputs' goes here.
// Store function startUpload
}).secret( 'in', 'startUpload', function(){
// 'this' points to the private object in $.secret
// where stores all the variables and functions
// ex. uploadCount, disableAllFileInputs, startUpload.
var fil = document.getElementById( 'FileUpload' + uploadCount);
if(!fil || fil.value.length == 0) {
alert( 'Finished!' );
document.forms[0].reset();
return;
}
// Use the stored disableAllFileInputs function
// or you can use $.secret( 'call', 'disableAllFileInputs' );
// it's the same thing.
this.disableAllFileInputs();
fil.disabled = false;
// this.uploadCount is equal to $.secret( 'out', 'uploadCount' );
alert( 'Uploading file ' + this.uploadCount );
document.forms[0].submit();
// Store function iframeOnload
}).secret( 'in', 'iframeOnload', function(){
this.uploadCount++;
if( this.uploadCount > 1 ) this.startUpload();
});
window.onload = function() {
var frm = document.forms[0];
frm.target = "postMe";
frm.onsubmit = function() {
// Call out startUpload function onsubmit
$.secret( 'call', 'startUpload' );
return false;
}
}
Relevant markup:
<iframe src="test.htm" name="postHere" id="postHere" onload="$.secret( 'call', 'iframeOnload' );"></iframe>
Open your Firebug, you will find no globals at all, not even the funciton :)
For full documentation, please see here.
For a demo page, please see this.
Source code on GitHub.
I use it this way:
{
var globalA = 100;
var globalB = 200;
var globalFunc = function() { ... }
let localA = 10;
let localB = 20;
let localFunc = function() { ... }
localFunc();
}
For all global scopes use 'var', and for local scopes use 'let'.
Use closures. Something like this gives you a scope other than global.
(function() {
// Your code here
var var1;
function f1() {
if(var1){...}
}
window.var_name = something; //<- if you have to have global var
window.glob_func = function(){...} //<- ...or global function
})();
For "securing" induvidual global variables:
function gInitUploadCount() {
var uploadCount = 0;
gGetUploadCount = function () {
return uploadCount;
}
gAddUploadCount= function () {
uploadCount +=1;
}
}
gInitUploadCount();
gAddUploadCount();
console.log("Upload counter = "+gGetUploadCount());
I'm a novice to JS, currently using this in one project.
(i apreciate any comment and criticism)
What happens is that within the function makeCounter:
function makeCounter() {
var i = 0;
return function() {
console.log( ++i );
};
}
You are returning a function, then to use it is the following:
const counter = makeCounter(); // execute function and return other function
counter(); // This executes the function that you returned
If for example you did not return a function, it would work as expected:
function makeCounter() {
var i = 0;
console.log( ++i );
}
makeCounter(); // this execute logs
One fun (YDOFMV1) way to avoid scope clutter for not just JavaScript but also HTML and even CSS(!) is with HTML templates. You do have to whip up a little architectural code (which of course doesn't have to be in global scope). Consider this HTML:
<html lang="en">
<head>
<title>Templates Are Fun!</title>
<script> // stuff!
</script>
</head>
<body>
<template>
<style> these style directives are local!</style>
<div> just about any html stuff! </div>
<script>
doStuff(); // this code ignored unless you eval it!
</script>
</template>
</body>
</html>
All the stuff in the template has no effect on your HTML page, so what good is it? First, the style element only affects HTML elements inside the template. Second, the script element is not actually evaluated, which means you can do that yourself and put the code inside a scope of your own choosing, with whatever environmental features you care to give it (all without messing about with "components" and whatnot).
On to the minor architecture you have to (get to!) do yourself:
For Each Template
Obviously, if you were doing a little single-page app, you could put each page in a separate template, then you swap in the rendered template as needed. Left as an exercise for the reader, but works great.
Also, this concept works great for recursive templates. I.e., the whole page is one big template, but it contains smaller templates (navigation bar, footer, etc.). Again, left as an exercise, but can't be that hard if I did it.
Whatever your scheme, you're going to need to get hold of that template node, set aside the script sub-node (so you can control its execution), and then replace the template node with the actual contents of the template. Your code may look at least vaguely like:
const template = document.querySelector('template');
const templateRoot = template.content;
const scriptNode = templateRoot.querySelector('script');
scriptNode.remove(); // take it out of the template
template.replaceWith(templateRoot); // replace template with its contents
After that code runs, the template contents now exist in your document and will no longer be ignored by the DOM parser. The script element still hasn't been executed but you have a reference to it in scriptNode.
Control That Script
You could eval the script, of course, but we're talking about limiting scope to avoid
problems
right? Use the Function constructor instead for greater control:
const code = '"use strict";' + scriptNode.innerHTML;
const templateFunc = new Function(code);
// now run templateFunc when you want
Note the injection of "use strict", because the code being evaluated does not inherit that setting from the calling code.
The code from the template script tag doesn't have access to much of anything other than the global scope, so you don't have to worry much about it colliding with any of your other code. Of course, that leads immediately to the problem that you may need that code have some interaction with the rest of your code.
Letting The Outside Call In
The "module" pattern works fine here. Since the template script will be wrapped in an anonymous function by the Function constructor, it can just return some interface that suits you:
// from inside template <script>
function init() { outside calls me to let me initialize };
function render() { outside calls me to tell me to render };
return {init:init, render:render};
Your world outside the template(s) then keeps that return value around to call into the template script code:
const code = '"use strict";' + scriptNode.innerHTML;
const templateFunc = new Function(code);
const templateInterface = templateFunc();
// let's tell the template script to initialize
templateInterface.init();
Letting the Inside Call Out
Besides the need to tell the template script what to do,
the template script probably needs some limited access to the outside
world. Again, we want to exercise total control over that access.
Once again, some flavor of "module" pattern works fine.
Suppose that we have an interface called "model" that contains data
some template scripts may need in order to render. Let's roll our own
require function, which we inject into the script code by making
require an argument to the anonymous function that the Function constructor creates:
// revised to give <script> code access to a 'require' function
// model defined elsewhere in this scope
function require(name){
switch(name){
case 'model' : return model;
// etc.
}
}
const code = '"use strict";' + scriptNode.innerHTML;
// anonymous function will have arg named "require"
const templateFunc = new Function("require", code);
const templateInterface = templateFunc(require);
// let's tell the template script to initialize
templateInterface.init();
Now the code inside the template script has access to
a variable (first argument of the anonymous function that encloses it)
named require that it can use in the standard ways:
<script>
const model = require('model');
// remaining code has access to model forever more
//...
<script>
Enriching the Environment
Of course, you can enrich the environment of the script code far beyond
just giving it a require function. This is why there are a billiondy frameworks for JavaScript. One I like is generating on the fly an object for each template that gives access to DOM elements it is interested in inside the template. This addresses the annoying problem that you often want to locate DOM elements by id, but then you have that nagging feeling that you're supposed to make all ids unique across the entire document, so id effectively becomes a global variable that you can't modularize.
Suppose that inside the template, you identify elements your script code cares about with a data-id tag:
<template>
<dialog id="PickOrCreateProject">
<center>
Project name: <input type="text" data-id="textProject" />
<input type="button" value="Create" data-id="createButton">
<br/>
<div data-id="div">
<select data-id="select"></select>
<input type="button" value="Select" data-id="selectButton">
</div>
</center>
</dialog>
<script> etc.</script>
</template>
And suppose that we want to give our script code easy
access to an object (call it viewNodes) that contains references to each node inside the template marked with a data-id attribute,
such that the script can do this:
const {textProject,createButton,selectButton} = viewNodes;
You can whip that up as easy as:
let viewNodes = {};
for(const element of templateRoot.querySelectorAll("[data-ID]"))
viewNodes[element.dataset.id] = element;
And revise the Function constructor to supply it to the
script code:
const templateFunc = new Function("require", "viewNodes", code);
const templateInterface = templateFunc(require, viewNodes);
Doesn't matter if two different templates use conflicting
data-id attributes, since no template ever looks outside
itself for the identifiers it wants.
Conclusion
In conclusion, templates are fun for modularity, and I must have some work I am
procrastinating hard on to waste time spewing this all out :-).
1Your Definition Of Fun May Vary
Related
I try to change some way to call methods into namespace.
Calling parent methods (I dont think its possible)
Creating and call inheritance function
Calling inside another method (mostly jquery onReady event function) (this.MyFunction() not working)
I split every namespace in files (want to keep it that way)
I try How to call function A from function B within the same namespace? but I didn't succed to split namespaces.
my fiddle sample got only 1 sub-namespace but could be more.
https://jsfiddle.net/forX/kv1w2rvc/
/**************************************************************************
// FILE Master.js
***************************************************************************/
if (!Master) var Master = {};
Master.Print= function(text){
console.log("master.Print :" + text);
$("body").append("<div>master.Print : " + text + "</div>");
}
/**************************************************************************
// FILE Master.Test1.js
***************************************************************************/
if (!Master) var Master = {};
if (!Master.Test1) Master.Test1 = {};
/**************************************************************************
* Descrition :
* Function for managing event load/documentReady
**************************************************************************/
Master.Test1.onReady = function () {
$(function () {
Master.Test1.Function1(); //try to replace because need all namespace.
try {
this.Function2(); //not working
}
catch(err) {
console.log("this.Function2 not working");
$("body").append("<div>this.Function2 not working</div>");
}
try {
this.Print("onReady"); //not working
}
catch(err) {
console.log("this.Print not working");
$("body").append("<div>this.Print not working</div>");
}
try {
Print("onReady"); //not working
}
catch(err) {
console.log("Print not working");
$("body").append("<div>Print not working</div>");
}
});
}
Master.Test1.Function1 = function () {
console.log("Function1");
$("body").append("<div>Function1</div>");
this.Function3(); //working because not inside another function
}
Master.Test1.Function2 = function () {
$("body").append("<div>Function2</div>");
console.log("Function2");
}
Master.Test1.Function3 = function () {
$("body").append("<div>Function3</div>");
console.log("Function3");
Master.Print("Function3"); //try to replace because need all namespace.
}
Master.Test1.onReady();
I use Master.Test1.Function1(); and I want to change that because Function1 is inside the same namespace.
I use Master.Print("Function3"); I dont think I can change that. the way I try to use it, it's more an inheritance function. but I dont know if theres a way to do that?
Maybe I should change the my namespace methode? maybe prototype will do what I want?
You can capture the this in a variable because this inside $(function() {}) will point to document object. The below will work provided you never change the calling context of onReady -- i.e. it is always called on the Test1 object and not called on other context:
Master.Test1.onReady = function () {
var self = this;
$(function () {
self.Function1();
// ..
});
}
To access Print you have to reference using the Master object like: Master.Print() as it won't be available in the Test1 object
this is document within .ready() or jQuery() alias for .ready() where function(){} is parameter $(function() {}). this at this.Function2() will reference document.
"Objects" in javascript are not built the same way as in most object-oriented languages. Essentially, what you are building is a hierarchy of static methods that have no real internal state in-and-of themselves. Therefore, when one of the defined methods is invoked, the context (or state) of that method depends on what object invoked the method.
If you want to have any internal context, you will need to create an "instance" of an "object prototype". At that point, you can use "this.otherFunction" within your other functions. Here is a small example:
var MyObject = function() {};
MyObject.functionOne = function() {
console.log("Function 1");
this.functionTwo();
};
MyObject.functionTwo = function() {
console.log("Function 2");
};
var instanceOne = new MyObject();
instanceOne.functionOne();
You might get some more information about object definition here
I've the following script:
gapi.analytics.ready(function() {
viewSelector.on('viewChange', function update (data) {
var title = document.getElementById('view-name');
title.innerHTML = data.property.name + ' (' + data.view.name + ')';
activeUsers.set(data).execute();
renderWeekOverWeekChart(data.ids);
renderTopBrowsersChart(data.ids);
renderTopCountriesChart(data.ids);
setTimeout(function() {
var list = document.getElementsByTagName("tr")[0];
list.getElementsByTagName("th")[0].innerHTML = "Pagina's";
list.getElementsByTagName("th")[1].innerHTML = "Paginaweergaven";
}, 500);
});
});
And within the following code I would like to re-run the update(); function.
function datumwissel( datumbtn ) {
if ( datumbtn.className == 'maand' ) {
datumbtn.className = 'jaar';
dimensions1 = 'ga:month,ga:nthMonth';
start1 = moment(now).date(1).month(0).format('YYYY-MM-DD');
end1 = moment(now).format('YYYY-MM-DD');
start2 = moment(now).subtract(1, 'year').date(1).month(0).format('YYYY-MM-DD');
end2 = moment(now).date(1).month(0).subtract(1, 'day').format('YYYY-MM-DD');
format1 = 'M';
format2 = 'MMM';
update();
}
else {
datumbtn.className = 'maand';
dimensions1 = 'ga:date,ga:nthWeek';
start1 = moment(now).subtract(2, 'day').date(1).format('YYYY-MM-DD');
end1 = moment(now).format('YYYY-MM-DD');
start2 = moment(now).subtract(2, 'day').date(1).subtract(1, 'month').format('YYYY-MM-DD');
end2 = moment(now).subtract(2, 'day').date(1).subtract(1, 'day').format('YYYY-MM-DD');
format1 = 'YYYYMMDD';
format2 = 'Do';
update();
}
}
But somehow this doesn't work. I also tried in the above script:
window.update = function (data) {}. But that also doesn't work.
How can I call the update(); function that is situated inside the gapi.analytics.ready(function() {} ?
Important is that I cannot make it globally as it has to be situated inside the gapi.analytics.ready().
It's really a simple matter of moving the function declaration
function update (data) {
// same as existing code
}
gapi.analytics.ready(function() {
viewSelector.on('viewChange', update );
});
And passing in data needed when you call it in your other function
function datumwissel( datumbtn ) {
if ( datumbtn.className == 'maand' ) {
..........
update(datumbtn);
}.......
Important is that I cannot make it globally as it has to be situated inside the gapi.analytics.ready()
That's not actually true - you can have it global and there at the same time. Whether you want to, is a different manner, as that would pollute the global namespace and so on. However, here is how that can be achieved:
First, extract the update function outside of the ready handler like so
function update (data) {
var title = document.getElementById('view-name');
title.innerHTML = data.property.name + ' (' + data.view.name + ')';
activeUsers.set(data).execute();
renderWeekOverWeekChart(data.ids);
renderTopBrowsersChart(data.ids);
renderTopCountriesChart(data.ids);
setTimeout(function() {
var list = document.getElementsByTagName("tr")[0];
list.getElementsByTagName("th")[0].innerHTML = "Pagina's";
list.getElementsByTagName("th")[1].innerHTML = "Paginaweergaven";
}, 500);
}
This will create a new function with the name update which accepts one parameter called data. Thanks to hoisting it would not matter if it's before or after anywhere you want to use it, as it would be effectively "pulled" to the top.
Next, you can just use the function inside the ready handler like so:
gapi.analytics.ready(function() {
viewSelector.on('viewChange', update);
});
Since .on(events, handler) accepts a function as the second parameter, you can just provide a function reference there. It doesn't matter that your function is technically declared elsewhere, as it is still going to be called with the same arguments. Similarly, if you replace update with alert you will be giving the reference to window.alert so you will get an alert with data.
With that, you can just call the same function in your other piece of code.
That is true for any place that uses callbacks, including setTimeout - you can just give a function reference and it's going to be called. Internally, those kinds of functions almost always do something like callback() or callback(someData), occasionally callback.call(/* parameters */) where callback is the passed in argument. Whether you define that argument as you are calling the function, e.g., selector.on("click", function() {/* code */}) or separately, e.g.,
function clickHandler() { /* code */ }
selector.on("click", clickHandler)
matters little.
With that said, whether you want the function global is a different matter. Unless both pieces of code are in the same place, a global function may be the easiest way. You could, also, namespace anything your app uses, which would partially avoid the global pollution. Not completely, but sometimes you just need to have things living under window if you have multiple files, in which case, you can define your own little corner there to play with: window.myApp = window.myApp || {} would create a new object that can serve as namespace and so you will be able to do things like myApp.update = function(data) { /* code */ } and thus share that code.
If your two pieces of code are indeed in one file, then you merely need to create the function outside both using var update = function(data) { /* code */ } then hand it to each in the exact same way, since update is still going to be a function reference, however, if assigned to a variable, it won't be added to the global namespace (nor would the declaration be hoisted).
I am relatively new to javascript and I am facing some difficulty.I have two java script files as I have shown below. I am having trouble getting the value of the variable entry_title inside the getRss function and storing it inside the variables Rss1_title and Rss2_title . Creating a global variable and assigning it to entry_title will make things worse as I will not be able to know from which Rss url the title came from. Is there a easy way to get the value of the callback functions ?
<script type="text/javascript" src="jsRss.js"></script>
<script type="text/javascript" src="notification.js"></script>
My notification.js file
function get_rss1_feeds(){
var Rss1_title = getRss("http://yofreesamples.com/category/free-coupons/feed/?type=rss");
}
function get_rss2_feeds(){
var Rss2_title = getRss("http://yofreesamples.com/category/real-freebies/feed/?type=rss");
}
setTimeout('get_rss1_feeds()',8000);
setTimeout('get_rss2_feeds()',7000);
My jsRss.js file:
function getRss(url){
if(url == null) return false;
google.load("feeds", "1");
// Our callback function, for when a feed is loaded.
function feedLoaded(result) {
if (!result.error) {
var entry = result.feed.entries[0];
var entry_title = entry.title; // need to get this value
}
}
function Load() {
// Create a feed instance that will grab feed.
var feed = new google.feeds.Feed(url);
// Calling load sends the request off. It requires a callback function.
feed.load(feedLoaded);
}
google.setOnLoadCallback(Load);
}
Errors :
When the setTimeout(get_rss1_feeds, 8000); method is called I get a blank screen.
I get a error in my console saying octal literals and octal escape sequences are deprecated and it is pointing to the 6th line in this script.
Is it because I am using google-api for parsing my Rss?
if (window['google'] != undefined && window['google']['loader'] != undefined) {
if (!window['google']['feeds']) {
window['google']['feeds'] = {};
google.feeds.Version = '1.0';
google.feeds.JSHash = '8992c0a2cdf258e5bd0f517c78243cd6';
google.feeds.LoadArgs = 'file\75feeds\46v\0751';
}
google.loader.writeLoadTag("css", google.loader.ServiceBase + "/api/feeds/1.0/8992c0a2cdf258e5bd0f517c78243cd6/default+en.css", false);
google.loader.writeLoadTag("script", google.loader.ServiceBase + "/api/feeds/1.0/8992c0a2cdf258e5bd0f517c78243cd6/default+en.I.js", false);
}
Seeing as it's a different scope, you can either return it in a callback, or provide it in another way such as exporting it to a higher scope that is visible to your desired location. In this case, it's the global scope, so I'd advise against that.
function getRss(url, callback) {
//...
function feedLoaded(result) {
if (!result.error) {
var entry = result.feed.entries[0];
var entry_title = entry.title; // need to get this value
callback && callback(entry_title);
}
}
and call it like so,
function get_rss1_feeds() {
var Rss1_title = getRss("http://yofreesamples.com/category/free-coupons/feed/?type=rss", function(entry_title) {
// This scope has access to entry_title
});
}
As an aside, use your setTimeout like so:
setTimeout(get_rss1_feeds, 8000);
rather than
setTimeout("get_rss1_feeds()", 8000);
as the latter uses eval, whereas the former passes a reference to the function.
Eventhough it will make your code a mess, you can append the variables to the window object.
For example:
function a()
{
window.testStr = "test";
}
function b()
{
alert(window.testStr);
}
Or even create your own object, instead of using window, as such:
var MyRSSReader = {
TitleOne : '',
TitleTwo : ''
}
MyRSSReader.TitleOne = "My title";
Wikipedia has a nice article about global variables, and why they are bad.
I have a working version of HTML5 drag & drop file uploader. I was editing the JS code to support multiple file uploads on same page. I came across with a problem by trying to access "instance" properties in methods which are registered as events.
The problem is shown by the code below, in method this.drop.
The reason of existence of the this.$$upload_self property is to access data through this property. For example, I can't use this keyword inside this.drop function, because when event is raised, this not referring my "instance".
I'm not sure that by creating $$upload_self was a good idea.
The new instances area created like this:
var recording_upload = new upload();
recording_upload.init(recording_upload, ...);
Code of Drag & drop file upload:
var upload = function() {
this.$$upload_self = null;
this.$drop = null;
this.$status = null;
this.$progress = null;
this.maxNumberOfFiles = null;
...
this.init = function (self, pUrl, maxNmbOfFiles, dropArea, progress, status) {
$$upload_self = self;
$$upload_self.postUrl = pUrl;
$$upload_self.maxNumberOfFiles = maxNmbOfFiles;
$$upload_self.$drop = $("#" + dropArea);
$$upload_self.$progress = $("#" + progress);
$$upload_self.$status = $("#" + status);
$$upload_self.$drop.bind('dragenter', $$upload_self.enter);
$$upload_self.$drop.bind('dragleave', $$upload_self.leave);
$$upload_self.$drop.bind('drop', $$upload_self.drop);
};
this.enter = function (e) {
$(e.target).addClass('hover');
return false;
};
this.leave = function (e) {
$(e.target).removeClass('hover');
return false;
};
this.drop = function (e, _this) {
$(e.target).removeClass('hover');
var files = e.originalEvent.dataTransfer.files;
if (files.length > $$upload_self.maxNumberOfFiles) { // for example, here $$upload_self references always last instance...
$$upload_self.displayErrorMessage('Error: You can only drop ' + $$upload_self.maxNumberOfFiles + ' file(s) at time.');
return false;
}
...
};
...
}
Is there any workaround to solve this issue? I believe this maybe a common problem, but can't find nothing to solve this problem.
Any help is very much appreciated.
You could ditch the new keyword altogether and use a fresh closure for each instance of upload.
EDIT: Updated to avoid potential clobbering of global this.
var upload = function(pUrl, maxNmbOfFiles, dropArea, progress, status) {
return {
postUrl: pUrl,
...
drop: function(e) {
...
if (files.length > this.maxNumberOfFiles) {
this.displayErrorMessage(...);
}
...
},
...
};
};
...
var someUpload = upload(...);
Try to search for a "scope". As an example see how it implemented in ExtJS.
In a modern browser you can do this:
$$upload_self.$drop.bind('dragleave', $$upload_self.leave.bind($$upload_self));
For older IE versions you can do this:
$$upload_self.$drop.bind('dragleave', function() { $$upload_self.leave(); });
Really the ".bind()" method of all Function objects in new browsers just creates a little intermediate function for you, essentially like the second example. The Mozilla docs page for ".bind()" has a very good block of code you can use as a "polyfill" patch for browsers that don't support ".bind()" natively.
I edited the question so it would make more sense.
I have a function that needs a couple arguments - let's call it fc(). I am passing that function as an argument through other functions (lets call them fa() and fb()). Each of the functions that fc() passes through add an argument to fc(). How do I pass fc() to each function without having to pass fc()'s arguments separately? Below is how I want it to work.
function fa(fc){
fc.myvar=something
fb(fc)
}
function fb(fc){
fc.myothervar=something
fc()
}
function fc(){
doessomething with myvar and myothervar
}
Below is how I do it now. As I add arguments, it's getting confusing because I have to add them to preceding function(s) as well. fb() and fc() get used elsewhere and I am loosing some flexibility.
function fa(fc){
myvar=something
fb(fc,myvar)
}
function fb(fc,myvar){
myothervar=something
fc(myvar,myothervar)
}
function fc(myvar,myothervar){
doessomething with myvar and myothervar
}
Thanks for your help
Edit 3 - The code
I updated my code using JimmyP's solution. I'd be interested in Jason Bunting's non-hack solution. Remember that each of these functions are also called from other functions and events.
From the HTML page
<input type="text" class="right" dynamicSelect="../selectLists/otherchargetype.aspx,null,calcSalesTax"/>
Set event handlers when section is loaded
function setDynamicSelectElements(oSet) {
/**************************************************************************************
* Sets the event handlers for inputs with dynamic selects
**************************************************************************************/
if (oSet.dynamicSelect) {
var ySelectArgs = oSet.dynamicSelect.split(',');
with (oSet) {
onkeyup = function() { findListItem(this); };
onclick = function() { selectList(ySelectArgs[0], ySelectArgs[1], ySelectArgs[2]) }
}
}
}
onclick event builds list
function selectList(sListName, sQuery, fnFollowing) {
/**************************************************************************************
* Build a dynamic select list and set each of the events for the table elements
**************************************************************************************/
if (fnFollowing) {
fnFollowing = eval(fnFollowing)//sent text function name, eval to a function
configureSelectList.clickEvent = fnFollowing
}
var oDiv = setDiv(sListName, sQuery, 'dynamicSelect', configureSelectList); //create the div in the right place
var oSelected = event.srcElement;
if (oSelected.value) findListItem(oSelected)//highlight the selected item
}
Create the list
function setDiv(sPageName, sQuery, sClassName, fnBeforeAppend) {
/**************************************************************************************
* Creates a div and places a page in it.
**************************************************************************************/
var oSelected = event.srcElement;
var sCursor = oSelected.style.cursor; //remember this for later
var coords = getElementCoords(oSelected);
var iBorder = makeNumeric(getStyle(oSelected, 'border-width'))
var oParent = oSelected.parentNode
if (!oParent.id) oParent.id = sAutoGenIdPrefix + randomNumber()//create an ID
var oDiv = document.getElementById(oParent.id + sWindowIdSuffix)//see if the div already exists
if (!oDiv) {//if not create it and set an id we can use to find it later
oDiv = document.createElement('DIV')
oDiv.id = oParent.id + sWindowIdSuffix//give the child an id so we can reference it later
oSelected.style.cursor = 'wait'//until the thing is loaded
oDiv.className = sClassName
oDiv.style.pixelLeft = coords.x + (iBorder * 2)
oDiv.style.pixelTop = (coords.y + coords.h + (iBorder * 2))
XmlHttpPage(sPageName, oDiv, sQuery)
if (fnBeforeAppend) {
fnBeforeAppend(oDiv)
}
oParent.appendChild(oDiv)
oSelected.style.cursor = ''//until the thing is loaded//once it's loaded, set the cursor back
oDiv.style.cursor = ''
}
return oDiv;
}
Position and size the list
function configureSelectList(oDiv, fnOnClick) {
/**************************************************************************************
* Build a dynamic select list and set each of the events for the table elements
* Created in one place and moved to another so that sizing based on the cell width can
* occur without being affected by stylesheet cascades
**************************************************************************************/
if(!fnOnClick) fnOnClick=configureSelectList.clickEvent
if (!oDiv) oDiv = configureSelectList.Container;
var oTable = getDecendant('TABLE', oDiv)
document.getElementsByTagName('TABLE')[0].rows[0].cells[0].appendChild(oDiv)//append to the doc so we are style free, then move it later
if (oTable) {
for (iRow = 0; iRow < oTable.rows.length; iRow++) {
var oRow = oTable.rows[iRow]
oRow.onmouseover = function() { highlightSelection(this) };
oRow.onmouseout = function() { highlightSelection(this) };
oRow.style.cursor = 'hand';
oRow.onclick = function() { closeSelectList(0); fnOnClick ? fnOnClick() : null };
oRow.cells[0].style.whiteSpace = 'nowrap'
}
} else {
//show some kind of error
}
oDiv.style.width = (oTable.offsetWidth + 20) + "px"; //no horiz scroll bars please
oTable.mouseout = function() { closeSelectList(500) };
if (oDiv.firstChild.offsetHeight < oDiv.offsetHeight) oDiv.style.height = oDiv.firstChild.offsetHeight//make sure the list is not too big for a few of items
}
Okay, so - where to start? :) Here is the partial function to begin with, you will need this (now and in the future, if you spend a lot of time hacking JavaScript):
function partial(func /*, 0..n args */) {
var args = Array.prototype.slice.call(arguments, 1);
return function() {
var allArguments = args.concat(Array.prototype.slice.call(arguments));
return func.apply(this, allArguments);
};
}
I see a lot of things about your code that make me cringe, but since I don't have time to really critique it, and you didn't ask for it, I will suggest the following if you want to rid yourself of the hack you are currently using, and a few other things:
The setDynamicSelectElements() function
In this function, you can change this line:
onclick = function() { selectList(ySelectArgs[0], ySelectArgs[1], ySelectArgs[2]) }
To this:
onclick = function() { selectList.apply(null, ySelectArgs); }
The selectList() function
In this function, you can get rid of this code where you are using eval - don't ever use eval unless you have a good reason to do so, it is very risky (go read up on it):
if (fnFollowing) {
fnFollowing = eval(fnFollowing)
configureSelectList.clickEvent = fnFollowing
}
And use this instead:
if(fnFollowing) {
fnFollowing = window[fnFollowing]; //this will find the function in the global scope
}
Then, change this line:
var oDiv = setDiv(sListName, sQuery, 'dynamicSelect', configureSelectList);
To this:
var oDiv = setDiv(sListName, sQuery, 'dynamicSelect', partial(configureSelectListAlternate, fnFollowing));
Now, in that code I provided, I have "configureSelectListAlternate" - that is a function that is the same as "configureSelectList" but has the parameters in the reverse order - if you can reverse the order of the parameters to "configureSelectList" instead, do that, otherwise here is my version:
function configureSelectListAlternate(fnOnClick, oDiv) {
configureSelectList(oDiv, fnOnClick);
}
The configureSelectList() function
In this function, you can eliminate this line:
if(!fnOnClick) fnOnClick=configureSelectList.clickEvent
That isn't needed any longer. Now, I see something I don't understand:
if (!oDiv) oDiv = configureSelectList.Container;
I didn't see you hook that Container property on in any of the other code. Unless you need this line, you should be able to get rid of it.
The setDiv() function can stay the same.
Not too exciting, but you get the idea - your code really could use some cleanup - are you avoiding the use of a library like jQuery or MochiKit for a good reason? It would make your life a lot easier...
A function's properties are not available as variables in the local scope. You must access them as properties. So, within 'fc' you could access 'myvar' in one of two ways:
// #1
arguments.callee.myvar;
// #2
fc.myvar;
Either's fine...
Try inheritance - by passing your whatever object as an argument, you gain access to whatever variables inside, like:
function Obj (iString) { // Base object
this.string = iString;
}
var myObj = new Obj ("text");
function InheritedObj (objInstance) { // Object with Obj vars
this.subObj = objInstance;
}
var myInheritedObj = new InheritedObj (myObj);
var myVar = myInheritedObj.subObj.string;
document.write (myVar);
subObj will take the form of myObj, so you can access the variables inside.
Maybe you are looking for Partial Function Application, or possibly currying?
Here is a quote from a blog post on the difference:
Where partial application takes a function and from it builds a function which takes fewer arguments, currying builds functions which take multiple arguments by composition of functions which each take a single argument.
If possible, it would help us help you if you could simplify your example and/or provide actual JS code instead of pseudocode.