W3C home > Mailing lists > Public > www-style@w3.org > May 2002

Re: canvas <html> <body>

From: <ewexler@stickdog.com>
Date: Sun, 5 May 2002 08:21:52 -0700
To: www-style@w3.org
Message-ID: <3CD4EBA0.10227.17F81C4@localhost>
George Lund wrote:

> Nothing in the (X)HTML specification defines how any given element must 
> be rendered (this is for optional style sheets to determine).
> 
> So whether <html> or <body> is taken as the root/canvas element for the 
> purpose of a particular rendering is dependent either on the style sheet 
> language and/or the implementation concerned.
> 
> The consensus appears to be that the <html> element can be the canvas 
> element for CSS.  (Subject to the user agent's default style sheet.)

CSS has a processing model and a rendering model that are clear on the issue of root 
elements.  The root element provides the root box, which fits in the initial containing 
block on the canvas.  The only question, really, is what element tree the CSS processor 
has at the start.  If the (X)HTML processor is defective, it may pass to the CSS 
processor a tree without an 'html' element at the root.  Or it may be that the user agent 
has an internal XSLT that legitimately transforms all XHTML documents to obtain a 
'body' root element.

> It would be bad practice (IMO) to style the <html> element or any of its 
> descendants (except <body>) because this would undermine the default 
> interpretation of (for example) <head> elements as providing *meta* 
> information.

I agree somewhat, but I differ on a few points.  I fail to see the harm in addressing 'html' 
elements for styling; after all, you want only 'head' elements to remain hidden.  But even 
there I cannot agree.  While I would not as a default display information in 'head' 
elements, that information can be helpful to a user and so I support its display at user 
option.  For example, a browser operating without a title bar (as in kiosk/presentation 
mode) could display the 'title' element.-- 
Etan Wexler <mailto:ewexler@stickdog.com>
Received on Sunday, 5 May 2002 11:36:05 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 27 April 2009 13:54:14 GMT