Side navigation
#12132 closed bug (fixed)
Opened July 24, 2012 01:42AM UTC
Closed July 24, 2012 06:53PM UTC
IE10 bug when cloning an object element without a parentNode
| Reported by: | elijahmanor | Owned by: | dmethvin |
|---|---|---|---|
| Priority: | blocker | Milestone: | 1.8 |
| Component: | manipulation | Version: | git |
| Keywords: | Cc: | ||
| Blocked by: | Blocking: |
Description
Only one of all the existing unit tests in jQuery fails when ran against IE10 http://cl.ly/image/0T1H3m1a0v1H
It is the following test in the manipulation module that is failing...
test("html() object element #10324", function() { ... });
Not too long ago a similar fix was added to workaround an issue with IE9 https://github.com/jquery/jquery/commit/2795a8390c1986200bf4e00158dbf3ad2da8d898
As it turns out IE10 throws a NoModificationAllowedError when setting the outerHTML of an element with a null parentNode. This behavior is different than IE9 and previous versions. I looked into the HTML5 spec and it says that the NO_MODIFICATION_ALLOWED_ERR error should be thrown if the parent is a Document http://html5.org/specs/dom-parsing.html#dom-element-outerhtml My guess is that IE10 took that new spec and took it a little further.
Attachments (0)
Change History (4)
Changed July 24, 2012 01:44AM UTC by comment:1
Changed July 24, 2012 01:58PM UTC by comment:2
| component: | unfiled → manipulation |
|---|---|
| milestone: | None → 1.8 |
| priority: | undecided → blocker |
| status: | new → open |
Confirmed. http://i.imgur.com/284Ld.png
Also looking into the long-async ready fail but that may be spurious.
Changed July 24, 2012 02:03PM UTC by comment:3
| owner: | → dmethvin |
|---|---|
| status: | open → assigned |
I had made a Pull Request to address this issue https://github.com/jquery/jquery/pull/870
As mentioned in the pull request...
In order to get around the issue, I am checking if the destination node has a parentNode before assigning to the outerHTML property. With this change IE10/9/8/7/6 all pass the current unit tests along with latest FF and Chrome releases. The code inside the cloneFixAttributes function that I modified is really to fix IE in the first place, so other browsers aren't really affected too much.
I initially also added a check to verify that the nodeType was not 9 (Node.DOCUMENT_NODE), but the code block that I changed only pertains to the object nodeName and I can't think of a use case where it's parentNode would ever be a document, so I think it is safe to leave out this check and leave the fix as terse as possible to save some bytes ;)
I may have been overgenerous with comments in this fix, but since it was an odd browser change I thought it might be helpful as I see other browser work arounds have similar comments.