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

Re: We are where we are.... but how did we get here?

From: L. David Baron <dbaron@dbaron.org>
Date: Wed, 29 Jun 2005 14:27:49 -0700
To: www-style@w3.org
Message-ID: <20050629212749.GA15458@ridley.dbaron.org>
On Wednesday 2005-06-29 13:43 -0400, Orion Adrian wrote:
> I was around the last time they were discussing the single versus
> multi-pass argument. Apparently multi-pass would complicate the
> algorithm. It's far from impossible, unless someone would like to
> correct me. It seems IE has a multi-pass system; what is it they've
> lost?

single versus multi-pass what?

If you're referring to the column styling issues and multi-pass selector
matching (which is what the responses to your message referred to), then
the answer is no, they don't have a multi-pass system.  They simply
don't implement the table values of the 'display' property, so the table
semantics in the markup always correspond to the rendering objects, and
they don't run into the problems described in
http://ln.hixie.ch/?start=1070385285&count=1 .  What they lose is the
ability to layout non-table markup using the table display types and the
ability to lay out table markup using other display types.

But I actually didn't see any reference to that in your message or what
it responded to.  It seems like you (and the message that you were
responding to) were referring to some multi-pass aspect of layout
algorithms, but it's not entirely clear what.  Do you mean making layout
depend more often on intrinsic widths?  (A common implementation of
intrinsic width calculation, which I suspect is used in IE and is used
in current Mozilla (although hopefully not for too much longer) is to
implement intrinsic width calculation in terms of a separate layout pass
at "infinite width".  However, it's not the only way to do it (I don't
think it's what Opera does and it's definitely not what KHTML and Safari
do) and probably not the best way, so I'd prefer not to refer to it as
multi-pass layout.)  If you do mean making layout depend more on
intrinsic widths (or heights, which generally doesn't require anything
resembling multiple passes), in what cases would you want the layout to
change?

-David

-- 
L. David Baron                                <URL: http://dbaron.org/ >
          Technical Lead, Layout & CSS, The Mozilla Foundation

Received on Wednesday, 29 June 2005 21:27:58 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 27 April 2009 13:54:38 GMT