W3C home > Mailing lists > Public > www-style@w3.org > January 2014

Re: [css3-box] How do scrollbars affect width calculation?

From: Thomas Rosenau <rosenau@seitenbau.com>
Date: Tue, 28 Jan 2014 10:36:33 +0100 (CET)
To: Gérard Talbot <www-style@gtalbot.org>
Cc: www-style@w3.org
Message-ID: <272948060.951208.1390901793333.JavaMail.zimbra@seitenbau.com>
Hello Gérard,

thanks for the clarification.

I realized that I accidentally didn't reply to the mailing list. Hope 
it is ok to share the message below.



> CSS width is to be substracted from the width of scrollbar even though
> the scrollbar is located between padding box and border box. So, content
> area is shrunk of "xxx" (from 9 "x" to 6 "x" in your ascii example)

Well, is it really the content area or just the containing block formed
by that element?

When using { background-clip: content-box; } it can be seen that the
content area is apparently not affected.
(cf. http://jsfiddle.net/LNMZU/)


Now, I found that the fundamental error I made was the assumption that a
box's text content would take the whole content width. However, this is
not the case, as can be seen from

http://www.w3.org/TR/CSS2/visuren.html#anonymous-block-level


So my current conjecture is that the following figure should be correct:

+-------------------------+-+
|         xxxxxx###       |^|
|         xxxxxx###       |~|
|         xxxxxx###       | |
|         xxxxxx###       | |
|         xxxxxx###       | |
|         xxxxxx###       |_|
|         xxxxxx###       |v|
+-------------------------+-+

| padding | width | padding |

          |    |
           width-scrollbarWidth

where xx=content and x#=content area.

Again, this is because the text content forms an anonymous block and
thus is constrained by the containing block. In other words: the
element's containing block does not coincide with the content area.
This feels a bit weird but it's the way the specs work.

I think we agree that computedStyle.width should not be influenced by
scrollbars.


Best regards,
Thomas

----- Ursprüngliche Mail -----
> Von: "Gérard Talbot" <www-style@gtalbot.org>
> An: "Thomas Rosenau" <rosenau@seitenbau.com>
> Gesendet: Montag, 27. Januar 2014 16:38:37
> Betreff: Re: [css3-box] How do scrollbars affect width calculation?
> 
> Le 2014-01-27 04:32, Thomas Rosenau a écrit :
> >> We already have several CSS 2.1 tests on this particular issue:
> > 
> > None of these cover the question how the content area is influenced.
> 
> Hello Thomas,
> 
> I merely replied to the "how so" question I saw. Getting computed style
> width of the element with a scrollbar should have returned 20px and not
> 5px or 3px in the example I read.
> 
> > 
> > I made a jsFiddle to illustrate the issue further:
> > http://jsfiddle.net/9pAcp/2/
> 
> 
> This is a much more elaborated example...interesting design.
> 
> clientWidth is very specific and must not be mixed up with css width.
> 
> I have this demo:
> http://www.gtalbot.org/BugzillaSection/DocumentAllDHTMLproperties.html
> 
> > You'll note how the content area's width (text wrapping) equals
> > cssWidth-scrollbarWidth,
> 
> Yes I see this. The content area is narrower. And this is what the spec
> says too.
> 
> i.e. what I observe is this:
> > 
> > +-------------------------+-+
> > |         xxxxxx          |^|
> > |         xxxxxx          |~|
> > |         xxxxxx          | |
> > |         xxxxxx          | |
> > |         xxxxxx          | |
> > |         xxxxxx          |_|
> > |         xxxxxx          |v|
> > +-------------------------+-+
> > 
> > | padding | width | padding |
> > 
> >           |    |
> >            width-scrollbarWidth
> > 
> > 
> > 
> > as opposed to this:
> > 
> > +-------------------------+-+
> > |         xxxxxxxxx       |^|
> > |         xxxxxxxxx       |~|
> > |         xxxxxxxxx       | |
> > |         xxxxxxxxx       | |
> > |         xxxxxxxxx       | |
> > |         xxxxxxxxx       |_|
> > |         xxxxxxxxx       |v|
> > +-------------------------+-+
> > 
> > | padding | width | padding |
> 
> 
> Yes. This is also what I observe. And this is what the spec states too.
> CSS width is to be substracted from the width of scrollbar even though
> the scrollbar is located between padding box and border box. So, content
> area is shrunk of "xxx" (from 9 "x" to 6 "x" in your ascii example)
> 
> > 
> > 
> > As I understand, this behaviour is consistent across browsers, but not
> > to be found in the specs.
> > 
> > 
> 
> It is found in the spec:
> 
> "
> In the case of a scrollbar being placed on an edge of the element's box,
> it should be inserted between the inner border edge and the outer
> padding edge. Any space taken up by the scrollbars should be taken out
> of (subtracted from the dimensions of) the containing block formed by
> the element with the scrollbars.
> "
> 11.1.1 Overflow: the 'overflow' property
> http://www.w3.org/TR/CSS21/visufx.html#overflow
> 
> and we have made CSS 2.1 tests in CSS 2.1 test suite about it:
> 
> http://test.csswg.org/suites/css2.1/nightly-unstable/html4/chapter-11.htm#s11.1.1
> 
> 
> > 
> >> The width of vertical scrollbar and the height of horizontal scrollbar
> >> are entirely user-settable by os settings.
> > 
> > I understand. The problem is: Firefox does never substract them, no
> > matter how wide they are.
> 
> Thomas, Firefox does not substract what from what?
> 
> Firefox 26 does not reduce computedStyle width of the scrollbar width
> because scrollbar is considered to be part of CSS width.
> 
> 
> 
> 
> That's because you mixed up content area (the bunch of "xxx" in your
> example) and computed style width.
> 
> CSS width is 9 "x" without a scrollbar or 6 "x" with a scrollbar: in
> either case, in both cases, the computedStyle CSS width should be the
> same.
> 
> The scrollbar is part of CSS width even though it may not be located
> inside it: that's tricky.
> 
> ----------
> 
>               <div style="width:20px;padding:20px;overflow:scroll"></div>
> 
> Right here, when I create a test, I try to use values which are unique
> and help noticeability, distinctiveness. So, here, I would choose width
> and padding values to be different.
> 
> <script>
> console.log(getComputedStyle(document.querySelector('div')).width);
> </script> Firefox returns '20px', while Chrome returns '5px' (with
> scrollbars
> 
>          being 15px wide)
> 
> GT: My opinion is that Firefox is correct since width of scrollbar is
> part of CSS width. Content area is narrowed by "xxx" in your ascii
> example and CSS width still remains 20px.
> 
> 
>              Which one is correct?
> 
>          Which Chrome? I get '20px' on Chrome 33.0.1750.51
> 
>      FWIW Chrome 32.0.1700.76 m returns '5px', 34.0.1807.0 canary '3px',
> IE 11 '3px' and Opera 18.0 also '3px', while Firefox 26.0 and 29.0a1
> (Nightly) return '20px'.
> 
> GT: My opinion is that there is a small bug with
> getComputedStyle(elem).width; Chrome 32, 34 and IE11 should return 20px.
> 
> 
> Gérard
> 
> > 
> > Best regards,
> > Thomas
> 
> 
Received on Tuesday, 28 January 2014 09:37:10 UTC

This archive was generated by hypermail 2.4.0 : Friday, 25 March 2022 10:08:39 UTC