CSS fallback: sensitive elements versus unsenstive ones

It feels to me that a large part of the discussion on @required
generally centers around the need for certain elements to work becuase
if they don't, the page is unusable. Perhaps as an academic exercise
it would be nice to identify the sorts of properties where this is a
problem and the sorts of properties where this isn't a problem.

My general conclusion after examining the various properties is that
properties that specify spacial layout cannot fail safely unless they
fail entirely; i.e. none of the properties in the set get applied.

Currently CSS allows for instance certain properties to fail without
affecting those properites that were specified at the same time. Now
while CSS does a fairly good job of specifying failsafe properties
inside a rule, it does not do such a good job when it comes to rules
for other elements. I may have a rule fail and now my interface is all
funky especially with any element that triggers on states like hover.

Even worse yet are properties that do something other than fail and
something other than succeed. Properties that do something inbetween
are usually fatal to any kind of spacial layout especially if you're
trying to produce something visually appealing. It doesn't take many
just slightly off lines to ruin the appearance of a document.

Now I've suggested that layout properties and behavior properties be
left for the UA to apply since it can know the capabilities of the
hardware much better than the author, but barring that from
consideration since it hasn't been well received, I ask this:

What besides @required (since it's untrustworthy) and besides pulling
layout out (since it isn't popular) and besides training (since it
takes lots of time and energy) can be done about creating safe
fallbacks for layout properties?

-- 

Orion Adrian

Received on Sunday, 18 September 2005 17:26:21 UTC