On Sat, 16 Jul 2005, David Woolley wrote:
> 
> Unfortunately there is a very strong lobby on both the CSS (relevant) 
> and SVG (not directly relevant) lists to have repeatable appearances 
> because browser developers find that authors and users are completely 
> incapable of understanding the concept of implementation choices and 
> therefore market forces force them to make their presentation alike.
>
> They therefore argue that if the specification doesn't mandate, they 
> would have to reverse engineer the market leader, making that the 
> specification, but not so well written a one.
> 
> I do agree that this is pandering to presentational markup and XHTML 2 
> really ought not to follow that line, even though CSS and SVG are 
> definitely following it at the moment.

There is a big difference between CSS and SVG (presentation languages) and 
HTML. CSS user agents (like most major Web browsers) have the concern you 
mention, namely wanting to ensure interoperable rendering. However, this 
should not extend to HTML-only UAs. There's no need for UAs that don't 
allow author-derived styling to have interoperable styling. (Thus, e.g., 
Opera in "user mode", especially on handheld devices, has a completely 
custom stylesheet.)

In short, I agree with what you say above, except that the point you make 
about specifications having to mandate doesn't extend to the rendering of 
HTML content by pure-HTML UAs.

(Mind you, the specification _does_ have to mandate error handling 
behaviour, for the exact same reason, and that _does_ extend to HTML. The 
sheer lack of any error handling logic in XHTML2 is of great concern.)

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