binding to beforeunload throws an error in IE6, 7, 8 on page unload

[info@…]

Take IE8 and start debug in script panel

1. Simply bind to the beforeunload event:

$(window).bind('beforeunload', $.noop);

2. Reload/leave the page

-> result:

SCRIPT5007: Unable to get value of the property 'type': object is null or undefined 
jquery.js, line 2206 character 5

A possible fix could look like this:

if ( !eventHandle ) {
	elemData.handle = eventHandle = function( e ) {
		// Handle the second event of a trigger and when
		// an event is called after a page has unloaded
		return typeof jQuery !== "undefined" && jQuery.event.triggered !== (e || window.event || {type: 'undef'}).type ?
			jQuery.event.handle.apply( eventHandle.elem, arguments ) :
			undefined;
	};
}

Diff:

Instead of:

e.type

do:

(e || window.event || {type: 'undef'}).type

This bug was introduced with jQuery 1.5.2 (don't know, why it was changed, so fix could look different)

[Rick Waldron]

From: ​http://api.jquery.com/bind/

"The .bind() method is the primary means of attaching behavior to a document. All JavaScript event types, such as focus, mouseover, and resize, are allowed for eventType. (The beforeunload and error events on the window object use nonstandard conventions and are not supported by jQuery; attach a handler directly to the window object instead.)"

[anonymous]

I'm now a little bit confused. jQuery does have code to handle 'beforeunload' event special (see $.event.special.beforeunload). Can you explain, why there is code for something, that is not supported?

[info@…]

Hi again,

I have changed my code, but it would be nice, if you could answer the question from the previous post.

Only, because I'm confused and want to know the answer. Am I missing something?

regards alex

[ajpiano]

​http://jsfiddle.net/p9tAp/1/show/ - jQuery 1.5.1 - works fine in IE ​http://jsfiddle.net/p9tAp/2/show/ - jQuery GIT (post 1.5.2) - broken in IE at present

It's my understanding that jQuery added real support for handling the beforenload event, and it was working fine in 1.5.1, so this is a regression and it should be fixed. It would appear to me that the documentation on .bind is currently incorrect.

[ajpiano]

#8737 is a duplicate of this ticket.

[Rick Waldron]

To explain your specific question above - I made a mistake, that's all :D

[addyosmani]

#8825 is a duplicate of this ticket.

[speednet]

I can confirm this bug, and I hope it's fixed soon. I patched jQuery with the fix mentioned in the initial post, and it solves the problem.

This jQuery regression bug breaks the Watermark plugin, because the Watermark plugin uses the beforeunload event to remove watermarks before form submission. (​http://jquery-watermark.googlecode.com/)

[John Resig]

Make sure that there isn't an error if no event object exists. Fixes #8755.

Changeset: adef5c3550141960089e2634619ef028931f4765

[addyosmani]

Updates to the docs related to this ticket have been made. Please see bind().