W3C home > Mailing lists > Public > www-style@w3.org > September 2006

Re: [css3-selectors] :parent selector

From: Ian Hickson <ian@hixie.ch>
Date: Thu, 28 Sep 2006 21:51:16 +0000 (UTC)
To: Steven Pemberton <steven.pemberton@cwi.nl>
Cc: David Hyatt <hyatt@apple.com>, Patrick Lauke <P.H.Lauke@salford.ac.uk>, www-style@w3.org
Message-ID: <Pine.LNX.4.62.0609282147310.4336@dhalsim.dreamhost.com>

On Thu, 28 Sep 2006, Steven Pemberton wrote:
> I think there needs to be a higher-level discussion, because there seem 
> to be some hidden assumptions in the current discussion that need to be 
> surfaced:
> 1. Are Selectors only for CSS, or are they intended to be used in other 
> contexts as well? If the latter, then deciding whether a selector is 
> needed shouldn't be based on grounds of rendering alone.

They're used, or are going to be used, in a number of contexts, the three 
main ones being rendering, XBL binding attachment, and DOM node selection. 
All three are highly performance sensitive.

> 2. Who is responsible for a page rendering incrementally: the CSS 
> designers, or the page author? If the latter, then the CSS specification 
> can simply mark certain selectors as obstructive to incremental 
> rendering, but still allow them for use.

The UA implementors. (Users demand incremental rendering of the browser 
vendors regardless of what the authors do.)

> 3. Does 'incremental rendering' mean rendering correctly immediately, or 
> does it mean showing the page as it loads, even if not in its final 
> form? Most browsers in my experience do the latter; for instance if the 
> stylesheet or embedded images are slow to load, then they will render as 
> best they can until they know more.

Actually most browsers try to avoid doing this -- IE for instance doesn't 
render tables until it knows their dimensions, Mozilla blocks on 
stylesheet loads, etc -- and the few browsers that do do this, e.g. 
Safari, suffer from well known issues that users find frustrating (e.g. 
the so-called "FOUC", or flicker of unstyled content), and which the 
browser vendors are actively trying to work around.

> But if 'incremental rendering' is code for 'doesn't need a DOM tree to 
> render' then this inherently limits the choice of selecters, whether the 
> result is rendered incrementally or not.

A DOM tree is required to handle modern-day content, if only for 

Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'
Received on Thursday, 28 September 2006 21:51:28 UTC

This archive was generated by hypermail 2.3.1 : Monday, 2 May 2016 14:27:26 UTC