I have sporadically been getting an "illegal access" exception in Chrome (29 and 30). Others have also seen this. It seems to happen on one specific line:
Here, this is an object I defined. It has a property end which is null or a number.
Does anyone know of anything that could cause an error with that message in Chrome?
EDIT: I don't expect anyone to debug ten thousands of lines of my code. What could cause an error with that message, whether it be jumping off the moon, singing blues, etc?
There are no occurrences of "illegal" in all of the Javascript code on the page.
This error has not happened in later versions of Chrome (31+).
Hopefully this is gone for good.
According to https://github.com/highcharts/highcharts/issues/2443, this would help:
try {
delete ret['e'];
} catch (e) {} // do nothing
We can't be sure, but depending on your network speed, hardware, your DOM wasn't totally and perfectly loaded.
You was probaby using recursion, many of them, with some hard javascript computation on each.
What happened is all tags and scripts of the DOM was already loaded and parsed, so the script started anyway. But content data not fully.
The javascript addEventListener(DOMContentLoaded", ) is more watching tags and their good closures to start his job, instead of the real full data load.
This behavior is possible on other browsers.
While your script was for sure too loud on his time, the javascript engine on all browsers had evolved enough to handle it smoothly, mostly on the memory part which was clearly the funnel.
Related
So, I'm writing an application that has ~5k lines of javascript... and it's stopped reporting errors. No firebug, IE triangle thing, nothing. It just won't work if there's an error. Anyone seen anything like this before? Debugging is somewhat... frustrating...
My best guess is too many layers of abstraction? But there aren't really that many.
Something else is wrong. It's not simply the size of your javascript. I'm running 70,000 lines of JavaScript in Chrome's debugger and FireFox 6's firebug and there are no debugging problems.
Look elsewhere. Is something not getting loaded? Do you need to clear cache and refresh? Try another browser (like Chrome) and see the results.
If it stopped reporting errors, its more likely those errors do not exist in your current revision, or that a block of code that you think is executing is actually never run. Set a breakpoint.
99% your script is not executing at all. You need manual trace, either with breakpoints, console.logs or alerts.
1% is that you have error handlers, who silence them. Look for suspicious try { .. } catch () {} and window.onerror.
Sometimes I get an error or an unhandled exception in a line in the jQuery UI js file. I know the problem was an empty or null object or property was passed to jQuery. For example in Chrome I get a 'Uncaught TypeError" error.
Doing some manual work I can find the culprit and I can do better error handling. This could take some time.
However I am looking for a fast automatic way to find the culprit by looking at the stacktrace when the exception occurred. Is there a modern browser which has this feature built in?
Or some JavaScript error handler which works across all the loaded js files in a global level?
I looked at this article but it seems I have to sprinkle printStackTrace() in all the targetted functions. I don't like this idea much if my code has many functions.
in webkit (what chrome or safari uses) debug tools, click the Scripts section on the top. then look for hexagon like icon on the bottom with the two vertical lines. clicking that will cause javascript execution to pause on an error. At the right you will see the callstack, where you can trace through everything.
I'm using Firebug to view JavaScript errors. But I'm also using jQuery. For some reason, I never see JavaScript errors when using jQuery and this is one of the biggest problems with working with jQuery in the first place.
There's no particular code to show, just imagine alert(areareg); where areareg is undefined or any sort of JavaScript error, and Firebug won't tell me about it. The JavaScript will simply fail without warning or notification of any sort. This is always the case, in any project where I've used jQuery and it's the only reason I don't like jQuery; because it's notoriously difficult to debug when something goes wrong.
For some reason I've even had trouble finding this question raised online, let alone answered. But I figured I'd give it a try here:
Is there any way to make jQuery stop suppressing error messages?
Am I the only one in the world who has always had this problem with jQuery?
EDIT:
I use both Firefox (with Firebug) and Chrome, and I only use the non-minified version of jQuery. Still, I have never in my entire life seen a jQuery error message of any kind, neither useful nor useless, nor in fact normal JavaScript errors, when using jQuery.
Well, despite your impression, jQuery itself doesn't do anything to "suppress" error messages. Now, Firefox does tend to throw away certain kinds of exceptions, notably like what you describe. I often find that wrapping a whole Javascript block in a try ... catch with an alert call should an exception happen is a useful way to deal with it, but it's still irritating.
I use google Chrome for my debugging, and it's pretty good, it shows all the jquery errors as well (as Alfred said, if you use minified version, you wont get any meaningful name in your errors, so it's better to use the raw source)
I am having the same problem in FF 18. I am using JSF and JQuery, though it happens even for plain JS code on the page. I intentionally put javascript code that should throw a syntax error without a word in Firebug. FF 18 has its own built-in console that looks similar to firebug. When I disabled the firebug plug-in the console.log messages started showing up. However, the errors are still not showing up. I just tried Chrome and the errors do appear there.
I've run into this problem several times. It's quite inconsistent and hard to replicate, sometimes it seems related to version issues. Settings rarely seem to make a difference, though "Break on next" sometimes (not always) identifies errors that otherwise slip through.
One thing I've noticed is, it tends to happen for code that is triggered inside a callback function, e.g. from a UI event or from an AJAX call. Sometimes Firefox / Firebug will behave like other browsers and show errors in the console, sometimes it gives nothing at all even for incredibly clear-cut cases like:
console.log('something'); // Shows up in console fine
fail // Other browsers give undefined variable error, Firefox gives nothing
console.log('something else'); // Never reached, so silent error clearly happens
The best approach I've found to this is:
Once you've found a no-show error, work back using console.trace() until you reach a point where inside a function, the error is hidden, outside, it shows, adding some code snippet like:
console.trace();
fail // cause the error, look out for a trace followed by nothing
If it's an AJAX call, manually add a basic error handler, because for some reason Firefox/Firebug is losing the default:
$.ajax({
error: function(a,b,c){
throw c;
},
other: '...settings'
}
If it's something you can't add an error handling function to, wrap it in a try... catch, for example:
try {
// whatever the code is within which there are no errors
} catch( err ){
throw err;
}
This should be enough to catch all errors where-ever they are within the scope where Firefox is losing them. Beware of cases where you have a callback within a callback though, you might need to do this twice...
I'm getting the following error in IE 6:
Line: 454
Char: 13
Error: Invalid Argument
Code: 0
URL: xxxxx/Iframe1.aspx
and I can't for the life of me find what's causing this.
This only happens in a situation where I have a main page that has several IFrames, and it only happens when I have one particular IFrame (the one pointed to by the URL in the error message), and that IFrame is invisible at the time of loading.
I've narrowed it up to there, but I still can't find anything more specific...
The IFrame in question doesn't have 454 lines in its HTML, nor do any of the JS files referred by it.
I tried attaching VS to iexplore.exe as a debugger, and it breaks when the error occurs, but then tells me "There is no source code available for the current location"...
Any suggestions on how I can go about chasing this one?
UPDATE: I found this problem through brute-force, basically, commenting everything out and uncommenting randomly...
But the question still stands: what is the rational way to find where the error is, when IE reports the wrong line number / file?
IE's Javascript engine is disgusting when it comes to debugging. You can try enabling script debugging in the Advanced Options, and then if you have Visual Studio installed it will jump to the place of error... if you're lucky. Other times you don't get anything, especially if the code was eval()'ed.
Another thing about these line numbers is that it doesn't reflect which file the error is happening in. I've had cases where the line number was actually correct, but it was in a linked .js file, not the main file.
Try using the Microsoft Script Debugger or DebugBar (http://www.debugbar.com) which may give you some better IE6 debugging tools. They always help me with IE6.
Also, does this happen in any newer versions of IE or just in IE6?
It's virtually impossible to debug this without a live example, but one thing that often causes an "Invalid Argument" error in Internet Explorer is trying to set an incorrect value for a style property.
So something like:
document.getElementById("foo").style.borderWidth = bar + "px";
when "bar" has the value null, or undefined, or is the string "grandma", will cause it, as "grandmapx" isn't a valid value for the borderWidth style property.
IE9 has a browser mode.
Open up Developer Tools, then select the version you want to emulate in the console, reload the page with errors, and the console will show you line numbers and the correct file where the error is.
I run into this problem a lot too, and I've also resorted to commenting everything out until I find the problem. One thing that I find to be useful is to add a try/catch block to every javascript method. Sometimes I add an alert to tell what method the error came from. Still tedious, but easier than trial and error commenting. And if you add them every time you write a new method it saves a lot of time in the event errors like those occur.
function TestMethod()
{
try
{
//whatever
}
catch (ex)
{
ShowError(ex.description);
//alert("TestMethod");
}
}
A note to other readers: I recently had this "Invalid argument." error reported in IE7-9 and eventually found that it was down to the way I was using setTimeout/setInterval.
This is wrong, in IE:
var Thing = {};
Thing.myFunc = function() { ... };
setTimeout(Thing.myFunc, 1000);
Instead, wrap the callback in an anonymous function like so:
var Thing = {};
Thing.myFunc = function() { ... };
setTimeout(function() { Thing.myFunc(); }, 1000);
and no more "invalid argument" errors.
Another possibility:
I do a lot of dev between two computers, at home and at work, so I often email myself or download pages from the server to work on. Recently I realised that Vista has a habit of unilaterally applying blocks to certain files when they are downloaded in certain ways, without notifying me that it is doing this.
The result is that, for example, an HTML page wants to access the .js file in its header, but it doesn't have permission to access local files. In this case, it doesn't matter what you write in the .js file, the browser will never even read it, and an irksome Line: 0 error will result.
So before you comb your code for an error, check your HTML page's properties, and see if it hasn't been blocked by the OS....
Like NickFitz pointed out, styling was an issue with my code.
document.getElementById('sums<%= event.id %>').style.border='1px solid #3b3b3b;'
I removed the border style and my IE issues were gone.
First of all, I've been researching this "Operation Aborted" error / bug for what seems like weeks, so here are a couple related questions and good articles if you are not familiar with it:
Why does ASP.NET cause the “Operation Aborted” Error in IE7? (so question)
Detecting cause of IE’s Operation Aborted Issue (so question)
Official Microsoft Knowledge base
Official IE Blog
Now here's my problem:
First I tried moving all my <script> elements to the end of my body tag. Didn't work. Then I refactored all my js functions to an external file that is linked in the <head>, all of my js functions are called from onclick or onkeypress anyway. Still getting the error. The last line of one of my .js files is
document.onload = setTimeout("foo()",500);
so I moved that to <body onload="setTimeout('foo()',500);">. I'm still getting this error. I don't know what to do. The only place I'm editing DOM elements is in foo(). Please help!
About my setup:
Java, Hibernate, Struts, JSPs ... I think that's all that is relevant.
What am I missing here?
Thanks in advance.
There are several causes for this. Two of the most common are:
1) scripts attempting to modify the DOM before the document has completely loaded
2) Trailing commas in object or array declarations
Number two is usually relatively easy to find, while number one is much tougher. Generally the best way to track down IE Javascript problems is to install Microsoft Script Debugger, so at least you can see what lines are causing the problem. With Script Debugger, IE will halt execution inside the browser and kick the script to a Script Debugger console, which will show the problem line. Much more informative than regular IE error messages.
Please see my answer to this question in another thread. I love this little trick and it has never failed me (when the cause is DOM manipulation before IE is ready, I mean). And as written, it doesn't affect the DOM-compliant browsers.
That problem can be a bear on a large page. Beyond the advice in the articles you already have, the only thing I can suggest from here is to remove wide swaths of the page in a dev environment until the problem goes away. Keep refining what is/is not on the page until you know which piece of content is causing the problem.
I've actually seen a confluence between two unrelated page elements cause this problem. I don't remember excisely why but the above approach, although painstaking, still worked.