W3C home > Mailing lists > Public > www-style@w3.org > July 2005

Re: Proposal: :column pseudo-class

From: Emrah BASKAYA <emrahbaskaya@hesido.com>
Date: Mon, 04 Jul 2005 14:10:39 +0300
To: "www-style.w3.org" <www-style@w3.org>
Message-ID: <op.stdyn1yl8nstxa@lomarnona>

On Sun, 03 Jul 2005 19:59:01 +0300, Laurens Holst  
<lholst@students.cs.uu.nl> wrote:

>> The cascade happens before layout. You don't know an element's display
>> type at the time you are doing selector matching.
> Fair enough... Never mind then :).

I have a question based on an assumption:

Assuming that the UA will know what coloumn the element is in as soon as  
it comes accross the elements opening tag during layout, can't it assign  
it an additional class automatically? Is this different than the UA not  
knowing which class an element is until actually it can actually load it?  
This coloumn scanning would only be done during layout and only for  
objects displayed as table elements, and would be done in the document  
appearance order, causing no-reflows. For cells appearing spanning  
multiple coloumns, the start coloumn class would be assigned. So, I'd  
think of it as assigning a new class based on the layout.

I don't -think- it would be a performance hit, to assess an elements  
coloumn during layout, and adding a 'virtual' class to it right before it  
is actually laid out with all the properties inherited by its 'virtual'  
parent, in this case, the coloumn selector. The performance required to  
know what element is in what coloumn -I think- may be negligible compared  
to the relatively monstrous task of actually laying the document out using  
CSS rules (floats with negative margins, margin collapsing, available  
space, etc etc.)

Of course, assumptions are not enough. How much performance hit it would  
be could be tested in real circumstances, and I certainly can't do it.

Received on Monday, 4 July 2005 11:11:05 UTC

This archive was generated by hypermail 2.3.1 : Monday, 2 May 2016 14:27:19 UTC