body.removeChild( container ); can cause an error

[Ian Yang]

​An answer of a question on stackoverflow shows that jQuery adds some elements (the outmost one is container) to do some tests (line 1537 in the un-minified version):

// Run tests that need a body at doc ready
jQuery(function() {
var container, outer, inner, table, td, offsetSupport,
...

And at line 1654 jQuery wants to remove that test element with:

body.removeChild( container );

However, as addressed in the question, by using the DOMNodeInserted event, all contents of body can be wrapped inside another element before jQuery can remove that test element, causing body.removeChild( container ); to fail because the container isn't a child element of body anymore.

So the suggestion is to change line 1654 into:

jQuery( container ).remove();

[timmywil]

DOMNodeInserted is deprecated and should no longer be used.

​http://www.w3.org/TR/DOM-Level-3-Events/#event-type-DOMNodeInserted

[dmethvin]

#12066 is a duplicate of this ticket.

[dmethvin]

Can you provide a test case on jsFiddle that reproduces the issue, not using the deprecated DOMNodeInserted?

[Ian Yang <[email protected]…>]

Replying to dmethvin:

Can you provide a test case on jsFiddle that reproduces the issue, not using the deprecated DOMNodeInserted?

​Here is the test case on jsFiddle. Please use the Firebug on Firefox 13 to see the error. And please click on the "Run" button to run it again if you don't see the error. On jsFiddle, the chance of this happening is 50%.

Note: the technique used in the test case is inspired by ​this article.

[gibson042]

Support loads before most other modules, in particular manipulation, so that line can execute before .remove is defined. And more importantly, we're just not prepared to tolerate external manipulation of our internal-use elements.

[dmethvin]

We could get around the error sans jQuery with container.parentNode.removeChild(container) but I don't think that resolves the issue -- it only hides it. If you are wrapping the HTML we use to do feature detects it is going to affect the jQuery.support outcomes, and cause jQuery to misbehave in strange ways down the road. The workaround would be do not attach your DOMNodeInserted handler until document ready, since we attach our ready handler for support checks first.

[Ian Yang <[email protected]…>]

Thanks for the advise anyway.

Btw, I have no intention of manipulating anyone's internal-use elements. I won't even know they exist if they didn't cause my browser to throw the error.