table type/class=layout (was: RE: [#293] Summary for tables)

> > 1. This will make people think that layout with tables is 
> > good, because  XHTML allows the value 'table'.

> Are tables (used for layout) really that bad...? I really see 
> Accessibility problems, but these would be removed 
> introducing the type  attribute. 

This overloads type, which may be worse than overloading
class.  Also note that an element can have several classes.

Other than that, I agree with you.  Tables for layout cause me
a fair amount of grief, because my screen (at my resolution)
typically isn't as large as the author expected.  On the other
hand, I hate to lose tables when they really are appropriate.

> > Do you think browsers not capable of rendering a CSS 
> > (CSS 2.0 rec is from 1998!) based layout will implement
> > the table type attribute earlier?

It can be done almost immediately; the real barrier is 
authoring tools.  Even use by hand-authors may be enough.
MSIE may (and probably should) take a while to release a 
new version, but the convention will be quickly supported 
where it is more important.

The default should be to continue displaying layout tables, 
just as the default is to use an author's CSS.  If desktop 
browsers don't have extra support, there is no harm.  

In environments where layout tables are a problem -- such 
as PDAs -- there are already clunky workarounds to turn 
off tables entirely.  This would let the user turn them off 
selectively, based on whether or not the particular table 
is really helpful on a small screen.

I will submit a patch to plucker (an offline web reader) the 
first time I notice this on even a single site that I regularly 
pluck.  It won't take much longer for proxomitron rulesets.   
I suspect that AvantGo, iSilo, and Opera's smallscreen mode
would also support it fairly quickly, if it starts to show up
in a fair number of pages.

-jJ

Received on Thursday, 17 July 2003 12:01:37 UTC