Re: Selector for parent/predecessor?

David Woolley <david@djwhome.demon.co.uk> wrote:

>> Using a single table to create a basic "layout" does not need to
>> increase the size of the code by more than a small and insignificant
>> amount when compared to so-called "CSS layouts".
>
>I suspect he may actually have been referring to the fact that the
>way that tables are used in the wild for layout prevents rendering
>on popular browsers until the whole table and all its subordinate
>resources have been transferred, preventing interaction with the
>page, and particular its first screenful, for many seconds on some
>connections.

That depends on the implementation, if an implementation allows reflows
if they are needed then the table can be rendered after the HTML has
been downloaded but dependencies have not.

Of course reflows are far from ideal, but the table-layout:auto
rendering model does provide what I consider an important quality for a
layout mechanism: the ability of grid cells to contain their content. 

The alternatives IMO create nastier problems than reflows:

"overflow:hidden" -> hiding content
"overflow:visible" -> obscuring content
"overflow:scroll" (or auto) -> the usability nightmare that in-document
CSS generated scroll bars form

-- 
Spartanicus

(email whitelist in use, non list-server mail will not be seen)

Received on Tuesday, 22 August 2006 10:02:14 UTC