:first selector behaves differently between IE8 and FF 3.6.8

[grimace]

I have the following html snippet:

<h1>Header</h1>
<div class="Item">Item 1</div>
<div class="Item">Item 2</div>

If I release the following selectors on it:

$(".Item:not(:first)")
$(".Item:not(.Item:first)")

I get Firefox to return 2 in both cases. IE8 will return 1 in the first case, and 2 in the second.

[dmethvin]

It seems like the correct answer would be 2 for the first case and 1 for the second. If everyone agrees, then IE8 gets it right in my test case and Firefox gets the first case wrong.

The :first element for Firefox 3.6 is the HTML element; in IE it's the !DOCTYPE comment element. In general it seems like a bad idea to use an unqualified :first selector but the Firefox behavior seems pretty strange.

[dmethvin]

Test case

[snover]

#5700 is a duplicate of this ticket.

[tomgrohl]

I've found that this works in Firefox 3.6.10, but not in IE8 ( but works fine for both when :last is used ):

alert( $(".Item:not(:first)").length );
alert( $(".Item:not(.Item:first)").length );

But this works in both ( returns 1 for length)

alert( $(".Item").not(":first").length );
alert( $(".Item").not(".Item:first").length );

[tomgrohl]

You gave me an idea when you said that $(":first") returns Html node for Firefox and Doctype comment for IE 8.

That would mean than the root (:root) would be different in each browser (well IE 8 anyways)

So I tried the following, adding body (html works too) in to the mix to give it a different root:

$("body .Item:not(:first)") //length of 1 for both
$("body .Item:not(.Item:first)") //length of 1 for both

And this works.

It seems if a different root is prepended to it, and not letting Sizzle set the root, then it works.

You can find a test of it ​here

Hope this is of some help.

[snover]

Retargeting due to severity of regressions in 1.4.3.

[snover]

Retargeting for 1.4.4 as per John’s request.

[snover]

Retargeting to next minor release.

[john]

I want to save this for 1.6 when the selector rewrite happens - a lot of this code will be changing and there's not much point in fiddling with it now.

[john]

Nominating ticket for 1.7 discussion.

[Rick Waldron]

+0, If it's a bug, it should be fixed, but this seems awkward

[jaubourg]

+0, Smells fishy... need a fiddle to see if document structure is the problem here.

[ajpiano]

+0, I guess that if it's a bug, then we should fix it, but also, there's a :gt selector for this...

[timmywil]

+1, sounds like a bug

[dmethvin]

-1, Let's push folks towards standards selectors and not muck around here.

[john]

+0, Ugh... :first semantics inside :not(), just terrible.

[scottgonzalez]

+0, There's clearly a bug in there and I wonder if it exists with other selectors too. However, I'm guessing this is really tricky to fix and there's an existing, better way to do the same thing.

[addyosmani]

+1

[jzaefferer]

+1