Re: More controlled degrading, @supported at-rule

On Sat, 18 Sep 2004, David Woolley wrote:
> 
> The reason that people keep proposing this sort of feature (my 
> preference is for an all or nothing block) is that even quite simple 
> designs can depend on multiple features being supported, whereas the 
> cascading rules only allow you to provide fallback on a per feature 
> basis.

Indeed; "co-dependent properties" is the what I called this some five or 
six years ago in this very list. The problem isn't only browser support, 
though, it's also user stylesheets and the cascade, device capibilities, 
etc. For example, this:

   header, footer { display: table-row; }
   navigation, content { display: table-cell; }

...(assuming markup where there are four elements in the <body>, namely 
<header>, <navigation>, <content>, and <footer> in that order) will work 
fine until the user decides to say

   header { display: table-cell; }

A more drastic example:

   body { background: black; color: white; }
   h1 { background: transparent; color: yellow; }

...with the user having said:

   body { background: yellow: color: black; }

There are lots of other examples of cases where it makes no sense for half 
the properties to cascade.

Note that none of these examples have anything to do with UA support.

Unfortunately there hasn't been a workable real solution to this problem.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

Received on Saturday, 18 September 2004 13:10:47 UTC