$(window).height() and .width() do not work in IE6 and IE7

[ebradsha]

I originally opened a ticket for this issue under #12300. However, it was prematurely closed out as 'worksforme' and I do not have the ability to re-open the bug.

Using $(window).width() and $(window).height() previously worked correctly with jQuery <1.8.0. In version 1.8.0, a value of 0 is always returned.

I have produced this bug under IE6 and IE7, I have not tested on other versions of IE.

I have tested this on multiple machines running IE6 and IE7 at multiple client sites with the same result every time. Unlike dmethvin, I do not have access to BrowserStack so I cannot speculate why the code worked on BrowserStack for him.

Here is a screenshot demonstrating the bug: ​http://s16.postimage.org/6zvnkd6r9/jquery_window_width_bug.png

Please let me know if you require any further demonstration of this bug to pursue debugging. I would be more than happy to grant a jQuery team member access to a VM running IE6 or IE7 so they can see this bug in action for themselves.

[dmethvin]

You weren't giving us complete information. Your test cases don't match what you have in the screen shot.

script defer is unfixably unreliable in IE. There's not even a reason to use it the way you do in the test case.

​https://github.com/h5bp/lazyweb-requests/issues/42

Can I close this now, or do you want to propose a solution of some kind after thoroughly reading the link above?

[ebradsha]

You can take out the defer, the bug remains. I put in the defer so the body text would load before the alert came up. Would you like me to post a new screenshot?

[dmethvin]

Please come up with a test case and stick to it.

[ebradsha]

Here is the new screenshot:

​http://s9.postimage.org/fuwrmbmgf/jquery_window_width_bug_no_defer.png

I'd be more than happy to grant you access to a dev VM running this test. Do you have a support/bugs email address I could send the login info to?

[dmethvin]

We have to be able to repro it, otherwise we can't create a unit test against it. To simplify recreating the problem and to be sure we're looking at the same code, can you post that page somewhere?

[ebradsha]

Here's an uploaded version, I added .height() in there as well:

​http://199.127.56.68/jquerybug12310.html

[ebradsha]

Screenshot of my results from the online version:

​http://s15.postimage.org/hg24bmrzv/jquery_bug_online_example.png

[nnyby]

I'm seeing this behavior in IE9 as well. Calling $(window).width() or $(window).height() with jQuery 1.8 returns 0. Using "document" instead of "window" works as expected, though.

[timmywil]

​jsbin test case

[anonymous]

This appears to have been broken by Pull Request ​https://github.com/jquery/jquery/pull/764 when a change was made to the dimensions.js file

There used to be a fallback check for the clientWidth off of the body, but that was taken out

Before Pull Request:

if ( jQuery.isWindow( elem ) ) {
    // 3rd condition allows Nokia support, as it supports the docElem prop but not CSS1Compat
    doc = elem.document;
    docElemProp = doc.documentElement[ clientProp ];
    return jQuery.support.boxModel && docElemProp ||
        doc.body && doc.body[ clientProp ] || docElemProp;
}

After Pull Request:

if ( jQuery.isWindow( elem ) ) {
    // As of 5/8/2012 this will yield incorrect results for Mobile Safari, but there
    // isn't a whole lot we can do. See pull request at this URL for discussion:
    // https://github.com/jquery/jquery/pull/764
    return elem.document.documentElement[ clientProp ];
}

If we add back in the fallback for the body check it fixes IE9. I've tested locally using the above JSBin that Timmy made with a tunnel to my localmachine with the updated code.

Proposed Slightly Reverted Code:

if ( jQuery.isWindow( elem ) ) {
    doc = elem.document;
    return doc.documentElement[ "client" + name ] || 
        doc.body && doc.body[ "client" + name ];
}

The trick is to know if this fix still addresses whatever was the intention of the Pull Request made 3 months ago. Seeing that how difficult this was to recreate the issue, I'm not sure how easy it will be to create a unit test showing that this fails inside our normal qunit test module.

I'll try to see if I can recreate the issue in a unit test within our existing qunit test suite and see if the above proposed fix addresses the original issue that was being addressed in the previous Pull Request.

[elijahmanor]

^{That was me by the way... didn't realize I wasn't logged into the system ;)}

[mikesherov]

​http://docs.jquery.com/Won%27t_Fix#Quirks_Mode

[mikesherov]

We need to redocument our lack of quirks mode support.

[mikesherov]

sorry, I missed the part where this wasn't quirks mode.

[mikesherov]

Actually, it is quirks mode!

​http://stackoverflow.com/questions/69828/must-doctype-be-the-very-first-tag-in-an-html-document

​http://jsbin.com/otipoq/11/

Apparently, jsbin renderes everything in Quirks Mode?!?

<script>setInterval=setTimeout;</script><!DOCTYPE html>

[mikesherov]

Remi just fixed JSBin, and the JSBin example no longer exhibits this issue.

​https://github.com/remy/jsbin/issues/312#issuecomment-7772785

Won't fix it stays.

[[email protected]…]

$(window).height() returns 0 in IE 10. for time being I am using $(document).height()