I seem to be observing at least one case where a callback function passed to a jQuery effect function will execute repeatedly if there's an error while it's executing.
For example, see this JS Fiddle, featuring the following code:
$('#awesome').fadeOut(400,function () {
log('fading out...');
dieInAFire();
});
log appends whatever's passed to it to a div... but dieInAFire doesn't exist. Rather simply stopping execution, however, the anonymous function appears to be getting called over and over, as evidenced by the growing number of appearances of 'fading out...' in the log div.
Is this the expected behavior? If so, why?
It's a known bug. See the report here.
I just submitted a comment on the bug that patrick dw posted.
Changing the line:
options.complete.call(elem);
To:
setTimeout(function(){
options.complete.call(elem);
}, 0);
Causes the callback to execute asynchronously, and if it will no longer stop execution if it throws any errors. IMO it's better than using a try catch since it doesn't suppress the exception.
If you want to edit your minified version, and you use the latest jQuery, you can search for e.complete.call(d) and replace it with setTimeout(function(){e.complete.call(d)},0)
Related
The Error in JavaScript internal/external file also stops the below code
For example:
var myObj = {};
myObj.other.getName = function(){
console.log('other is not defined');
};
alert('this will not show');
in above code, the alert will not come as the above code has error.
I added the same code in one file tst1.js and below this add one more file tst2.js. put alert('in tst2.js') in it. the tst2 alert come while tst1 not. it is some what related to
code compilation/interpretation.
It's much appreciated If someone explain me this behavior :)
This is the default behaviour of JavaScript. Avoid errors and the code will run normally.
Also you can handle errors with try...catch: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Statements/try...catch
All JS implementations are, AFAIK, single threaded. This means that all of your code is executed sequentially, on a single thread (logically). If this thread encounters a fatal error, it grinds to a halt. All code that comes after the point where the error occurs is ignored (the JS engine halted, no work is done anymore).
To clarify: it does not matter how many files you have. All of the JS code is stringed together into one big script, and this one script is executed sequentially (execution point starts at line 1 of the first script, and ends at the last line of the last script). Any errors in that code will cause the overall execution to grind to a halt:
//file1.js
var foo = (function()
{
console.log('This file is flawless, but pointless');
}());
//file2.js
foo();//calls previously defined function, assigned to var foo
//file3.js
fudhfsiufhi;//ERROR
//file4.js
foo();//will never get executed, because an error occurred in file3.js
Remove file3, or indeed fix the error, and everything will work as expected.
Allthough JS code is executed/evaluated sequentially, event handlers, callbacks, intervals, and timeouts might lead you to believe otherwise. Couple that with the fact that you have some control over what code is executed when, but not full control, and you get situations that, at first, seem rather counter intuitive. Consider this:
setTimeout(function()
{
massive(syntaxError) -123 + "here";
},0);//zero MS
alert('This will show');
This oddity is well documented, but it has to do with JS having a callback/handler loop, and queue. The setTimeout sets a timeout, to call the anonymous function in 0ms (immediately), but that callback is sent to the queue, which is checked periodically. Before the queue is checked (and the callback invoked), the alert will show. That's why the interval you pass to setTimeout or setInterval is not guaranteed to be exactly N milliseconds.
You can postpone a call to a method somewhat, by adding a call to the queue, like in the snippet above. But when the queue is processed, and what order the queued calls will be performed in are things you have no real say in. No say whatsoever.
It doesn't matter how many files, or how many statements that come before or after the problematic piece of code: there is no thread left to carry on.
The code you posted has a pretty clear error in it: you're assigning a property to other (a property of myObj, but this property is not defined anywhere, let alone defined as an object. Fix it by declaring properties first, before accessing them:
var myObj = {other: {}};//other is an empty object literal
myObj.other.getName = function()
{
console.log('This works');
};
alert('And the alert will show');
Your current code evalutes to:
var myObj = {};//object
var (myObj.other).getName = ...;
//evaluates to undefined
undefined.getName = ...//ERROR
undefined is a primitive value, actually signifying the absence of a value. undefined, therefore, cannot be accessed as an object, it can't be assigned properties.
Note:
This is just for completeness' sake, but JS is indeed single-threaded most of the time, but ECMAScript5 introduced Worker's which allow for some restricted form of multi-threading (without shared state, for example). Read through the MDN documentation on workers if you want to know more.
I have what looks like to me to be a simple variable assignment not working.
This code is in jQuery, for the context see here.
I'm calling:
$('#foo').on('someEvent', eventHandlerFn);
And I get this issue within the jQuery on function. Here's the starting point:
As you can see from the console below the code, selector is set the my eventHandlerFn and the fn variable is undefined. This is as expected.
On line 3509, the value of selector is assinged to fn. So, the value of fn should be same as the value of selector, no??
See below - selector is defined, as expected, but fn is still undefined. Why?
The end result is that my event handler is never registered.
The code runs well as shown in the following two screens (the issue is on how chrome sets the context to the console)
It looks like console has access to the variable at definition time (in this case the passed parameters) and not the live values as you run the code
Before the swap
After the swap
I'm not seeing any problem with this jsFiddle. Feel free to edit the jsFiddle to get it to look more like your code.
Can you try putting in console.log(fn); after line 3510 and rerunning? Maybe it's just a problem with the debugger?
This seems to be an issue with the debugger in Chrome - either a material problem or just a nuance of the debugger that I don't understand. fn does have a value toward the end of the call, but not where the breakpoint is.
I'm updating an existing website running on Expression Engine. So far, I've stayed away from any code I didn't write or couldn't understand. I recently must have altered some bit of code someplace (helpful, I know) and now a block of JS I didn't write is causing an error that seems to bypass the document.ready() event. The window.load() event however is still taking place.
In the Chrome DevTools Console, the error "Uncought TypeError: Cannot call method 'replace' of UNDEFINED" points to the definition of a function "fixedEncodeURIComponent" pasted below.
$("#MessageContainer.Counted").counter({
type: 'char',
goal: 250,
count: 'down'
}).change(function(){
var TEMP = fixedEncodeURIComponent($(this).val());
$("#Message").val(TEMP);
});
var TEMP = fixedEncodeURIComponent($("#MessageContainer.Test").val());
$("#Message").val(TEMP);
function fixedEncodeURIComponent (str) {
str=str.replace(/"/g, '');
return encodeURIComponent(str).replace(/[!'()*]/g, escape);
}
As I interpret the error, this function is being passed a variable that is not a string. I added an alert(str) to the function definition and the result was UNDEFINED as I expected. The first of several unknowns for me is which call to the function 'fixedEncodeURIComponent' is being passed a bad variable. I assume that it's the first call, but that's just a guess. It so happens that this first call contains a syntax I have never encountered before. I don't know how to interpret what happens when $(this) is passed as a function argument.
Any insights would be greatly appreciated. Also, if there's more information you need please let me know. The client's site is password protected but I can include any code you request.
Thank you.
I'm taking a guess that the }); on line 3 is exiting a document.ready context. If that's the case then your second call to fixedEncodeURIComponent may be getting called before the DOM is even loaded.
Start by wrapping
var TEMP = fixedEncodeURIComponent($("#MessageContainer.Test").val());
$("#Message").val(TEMP);
in a
$(function() {
// code
});
block. If that doesn't work, check that #MessageContainer.Test actually matches an element. Since this is code you inherited, the class name "Test" clues me in that the block in question might be a remnant of someone trying to debug an issue and maybe it should have been removed.
I suspect $("#MessageContainer.Test") since it looks like its supposed to be an ID selector instead of what it actually is when jQUery parses it(which is an ID selector combined with a class selector). $("MessageContainer\\.Test") allows you to select an element with ID MessageContainer.Test
For the following code in Javavscript:
// add HTML to container
// show the container on the DOM
// call a non-existent function on an object
I successfully see Firefox and IE display my HTML. The third line of code, which shows a JS error in Firebug and IE, appears to be suppressed/ignored.
Do browsers generally ignore bad JS? Or, let's say I added alert("line 4"); after my 3rd line of code. Would the 4th line be ignored since JS interpretation would end on the illegal JS line of code?
I tested my scenario in jsFiddle, but I'm not sure if jsFiddle behaves in the same way as a browser.
The JavaScript engine will execute code until it reaches an Exception. The behaviour then changes depending on the following:
If it is in a try..catch, execution will resume from catch, otherwise
If it is invoked asynchronously this sequence will end at that point but other ongoing ones will continue. (A simple example is with window.setTimeout)
If it is directly in a <script>, the rest of the code in the <script> from that point will not execute, but code in following <script>s will.
If the Exception occurs in something that gets hoisted, then think of the point of execution as being on line 0.
You can test what happens easily using the throw keyword. For example
console.log(1); // logs
console.log(2); // logs
throw 'eep';
console.log(4); // does not log
I have a pretty simple thing I'm doing with javascript and basically only sometimes will javascript give me a "too much recursion" error.
The code in question:
if(pageLoad===undefined){
var pageLoad=function(){};
}
var pageLoad_uniqueid_11=pageLoad;
var pageLoad=function(){
pageLoad_uniqueid_11();
pageLoad_uniqueid_12();
};
var pageLoad_uniqueid_12=function(){
alert('pageLoad');
};
$(document).ready(function(){
pageLoad();
});
(yes I know there are better way of doing this. This is difficult to change though, especially because of ASP.Net partial postbacks which aren't shown).
Anyway, when the too much recursion error happens, it continues to happen until I restart Firefox. when I restart Firefox it all works as normal again. How do I fix this?
I've also made a jsbin example
Update
Ok I've found out how to reliably reproduce it in our code, but it doesn't work for the jsbin example. If I create a new tab and go to the same page(have two tabs of the same address) and then refresh the first tab two times then I get this error consistently. We are not using any kind of session or anything else that I can think of that could cause such a problem to only occur in one tab!
Update 2
Not as reliable as I thought, but it definitely only occurs when more than one tab of the same page is open. It'll occur every few reloads of one of the tabs open
I've also updated my code to show an alert when pageLoad(the if statement) is initially undefined and when it is initially defined. Somehow, both alerts are showing up. This code is not duplicated in the rendered page and there is no way that it is being called twice. It is in a top level script element not surrounded by a function or anything! My code ends up looking like
if(pageLoad===undefined){
var pageLoad=function(){};
alert('new');
} else {
alert('old');
}
The code in question -- by itself -- should never result in an infinite recursion issue -- there is no function-statement and all the function objects are eagerly assigned to the variables. (If pageload is first undefined it will be assigned a No-Operation function, see next section.)
I suspect there is additional code/events that is triggering the behavior. One thing that may cause it is if the script/code is triggered twice during a page lifetime. The 2nd time pageload will not be undefined and will keep the original value, which if it is the function that calls the other two functions, will lead to infinite recursion.
I would recommend cleaning up the approach -- and having any issues caused by the complications just disappear ;-) What is the desired intent?
Happy coding.
This is just some additional info for other people trying to look for similar "too much recursion" errors in their code. Looks like firefox (as an example) gets too much recursion at about 6500 stack frames deep in this example: function moose(n){if(n%100 === 0)console.log(n);moose(n+1)};moose(0) . Similar examples can see depths of between 5000 and 7000. Not sure what the determining factors are, but it seems the number of parameters in the function drastically decrease the stack frame depth at which you get a "too much recursion" error. For example, this only gets to 3100:
function moose(n,m,a,s,d,fg,g,q,w,r,t,y,u,i,d){if(n%100 === 0)console.log(n);moose(n+1)};moose(0)
If you want to get around this, you can use setTimeout to schedule iterations to continue from the scheduler (which resets the stack). This obviously only works if you don't need to return something from the call:
function recurse(n) {
if(n%100 === 0)
setTimeout(function() {
recurse(n+1)
},0)
else
recurse(n+1)
}
Proper tail calls in ECMAScript 6 will solve the problem for some cases where you do need to return something from calls like this. Until then, for cases with deep recursion, the only answers are using either iteration, or the setTimeout method I mentioned.
I came across this error. The scenario in my case was different. The culprit code was something like this (which is simple concatenation recessively)
while(row)
{
string_a .= row['name'];
}
I found that JavaScript throws error on 180th recursion. Up till 179 loop, the code runs fine.
The behaviors in Safaris is exactly the same, except that the error it shows is "RangeError: Maximum call stack size exceeded." It throws this error on 180 recursion as well.
Although this is not related to function call but it might help somebody who are stuck with it.
Afaik, this error can also appear if you state a wrong parameter for your ajax request, like
$.getJSON('get.php',{cmd:"1", elem:$('#elem')},function(data) { // ... }
Which then should be
elem:$('#elem').val()
instead.
This will also cause the "too much recursion" issue:
class account {
constructor() {
this.balance = 0; // <-- property: balance
}
set balance( amount ) { // <-- set function is the same name as the property.
this.balance = amount; // <-- AND property: balance (unintended recursion here)
}
}
var acc = new account();
Using unique names is important.
Ok, so why is this happening?
In the set function it isn't actually setting the property to amount, instead it's calling the set function again because in the scope of the set function it is the same syntax for both setting the property AND calling the set function.
Because in that scope this is the same as account and (account OR this).balance = amount can both call the set function OR set the property.
The solution to this is to simply change the name of either the property or the set function in any way (and of course update the rest of the code accordingly).