outerWidth(true) and css('margin') returning % instead of px in Webkit

[Sprenger]

If you have a relative margin (defined in %), jQuery(element).css('margin-right') will normally return the margin value in pixels not in percentage. In Webkit, this is not the case.

The method jQuery(element).outerWidth returns in IOS5 the addition of width (pixel) and margin (percentage) which mixes two different measures.

Test file:  http://www.chateau-web.ch/demo/ios5-jquery-2.html

Tested with 1.6.4 and v1.7 (git) and 1.5. (same values for all versions)

[timmywil]

Confirmed. Probably related to the effects percentage issue.

Test case ported to jsfiddle:  http://jsfiddle.net/timmywil/u4F8m/

[mikesherov]

chrome and FF are fundamentally different in how they report computed values. :-\

[mikesherov]

Unfortunately, it looks like a "wait for webkit to fix this" situation. The hack in the provided fiddle assumes that the parent is a pixel width element. In order to correctly work around the chrome bug, we would have to perform a russian doll approach, continuing to go up the tree correcting and coenverting all the way up the DOM, possibly until we're measuring the document's width/height until we get a sane value. I'm not sure jQuery wants to support that, but that's just my two cents.

Here are some resources related to this bug.

• Currently open bug in webkit regarding using "computed values" instead of "used value":  https://bugs.webkit.org/show_bug.cgi?id=29084
• used value article from MDN:  https://developer.mozilla.org/en/CSS/used_value
• YUI bug report for the same thing that is essentially a tumbleweed:  http://yuilibrary.com/projects/yui3/ticket/2529799

[mikesherov]

I advocated for the bug on the chrome bugzilla:  https://bugs.webkit.org/show_bug.cgi?id=29084

Anyone else feel like pitching in an opinion?

[mikesherov]

Gaining some traction to be fixed in webkit:  https://bugs.webkit.org/show_bug.cgi?id=29084

[mikesherov]

found a better hack... still working out the kinks.

[mikesherov]

 https://github.com/jquery/jquery/pull/616 Still some potential perf issues to work out.

[mikesherov]

By the way, as an aside, open this fiddle in IE9,FF,0< or Webkit, they all report different values. getComputedStyle() and margins aren't friends:  http://jsfiddle.net/u4F8m/13/

[mikesherov]

I've updated the PR, still pretty slow when this situation is encountered, but fast otherwise:  https://github.com/jquery/jquery/pull/616

[mikesherov]

#10117 is a duplicate of this ticket.

[mikesherov]

Just an update, this is going to be fixed in chrome itself, but this hack may still be necessary for older safari.  https://twitter.com/#!/jarrednicholls/status/142049583915204608  https://bugs.webkit.org/show_bug.cgi?id=73334

[Mike Sherov]

Fix #10639. Make percent-specified margins return px values in WebKit.

Changeset: 7f6a991313380b74d5fb18782fb6b99fd6c4a22d

[heygrady]

I've left a  comment on the commit on GitHub because I'm not allowed to comment here. Fixing percentages in WebKit could be more robust.

[ntnwrd@…]

Strangely, if an element has css 'min-width' value and a css margin value that is a %, in Safari (Version 5.1.7 (7534.57.2)), the returned value for $(element) is the 'min-width' value!

 http://jsfiddle.net/E2U5C/6/

[mikesherov]

ntnwrd@…, please submit a new bug for that as its a different bug, please.