[Webkit-unassigned] [Bug 13343] getComputedStyle returns wrong value for margin-right

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=13343





--- Comment #23 from Jarred Nicholls <jarred.nicholls at gmail.com>  2011-02-04 08:53:32 PST ---
Bare with me, this will be a lengthy comment, but I just wanted to weigh in my opinion on this bug.

My feeling is that the original behavior is the correct one, OR the current behavior of width/height (among others) is wrong.  We can use CSS 2.1 (http://www.w3.org/TR/CSS21/visudet.html#the-width-property & http://www.w3.org/TR/CSS21/box.html#margin-properties) or CSS 3 (http://www.w3.org/TR/css3-box/) as our guide.

Margin's computed values are said to be this: "the percentage as specified or the absolute length or ‘auto’".  Some have understood "absolute length" to be what was specified (including Firefox, Opera, IE, and everyone CC'ed on this bug).  The original behavior understood "absolute length" to mean "the used value".

Now if we take width/height's computed value definition: "the percentage or ‘auto’ as specified or the absolute length; ‘auto’ if the property does not apply".  You'll notice that "absolute length" is again used, when a percentage or auto isn't specified.  The difference here is that all browsers, including WebKit, always return "the used value" as the computed width.  So if we specified "auto", "50%", or "500px", we always get the used value in fixed pixels back.  If our container block is 1000px wide, the computed value for those 3 cases would be "1000px", "500px", and "500px" respectively.

So currently, margin-left: auto; and margin-right: auto; will return a computed value of the "used value" in fixed pixels.  If this isn't right, then width/height aren't right either.

So the question is, how do we define "absolute length".  Is it the absolute fixed length specified, or is it the used value?  It seems like the world has come to a conclusion that it's the specified value.  In the case margin is not specified, an initial value of 0 is used.  In the case of width/height though, it would appear that no browser is compliant.

I would expect width: auto; to return "auto", width: 50%; to return "50%", and width: 500px; to return "500px", and an undefined width to return "auto".  The CSS3 patch I've attached to this bug will make margin's compliant to CSS3, but it seems width/height (and several others) need fixing too.

So what does everyone think?  Do we go full compliant, or do we just try to act like the rest of the browsers?  Thanks for reading my rant :)

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.