[_] Neilsen & Dvorak (was: New Microsoft web site design)
Tim Beadle
tim.beadle at gmail.com
Wed Jul 26 09:59:36 BST 2006
On 26/07/06, Chris Kaminski <chris at setmajer.com> wrote:
> But one of his two main points is the really difficulty of the cascade. He's
> not wrong about that: how many times have you been handed a site with a CSS
> bug and had to chase up the cascade looking for the rule that's causing the
> problem?
Fair point - "View Style Information" to the rescue. It's made life so
much easier.
> I do it all the time, and while tools like Firebug and Dreamweaver's tag
> inspector help they're not infallible -- add a bit of scripting or (in the
> case of DW) a CMS and they quickly become inaccurate.
We don't have too much scripting ATM, so it's been OK up to now.
> That said, I'm not at all certain that eliminating the cascade would help
> any -- having to define font styles for every element, for example, would
> get very tedious very quickly and send a whole lot of extraneous crap down
> the pipe. One could potentially mitigate the problem with an entity/macro
> system, but that would only cut down on the characters sent, not on the
> complexity of adding the appropriate styles to all possible elements --
> particularly for sites where you'll be displaying user-generated content.
>
> So I'm not sure there's a good solution to the problem -- though I'd love to
> see someone try to find one.
Maybe I've been too close to CSS for too long, but conceptually it's a
very elegant solution. The implementation might suck sometimes, but
that keeps life interesting.
> His other point is that CSS is 'broken' because each browser interprets it
> differently. That's a yes-and-no situation. It's true that the CSS standard
> isn't precise enough to expect exact agreement from all user agents -- but
> that's a necessity given the number of environments in which CSS formatting
> will be displayed.
>
> OTOH, even if CSS 2.1 *were* so precise and bug-free that it could be
> implemented perfectly by every UA, there's no way to force UAs to actually
> implement it completely and without error. And it's browser bugs and
> incomplete support that are the root of 99% of the trouble -- and those have
> squat to do with 'standards organisations'.
Indeed.
Tim
> But one of his two main points is the really difficulty of the cascade. He's
> not wrong about that: how many times have you been handed a site with a CSS
> bug and had to chase up the cascade looking for the rule that's causing the
> problem?
Fair point - "View Style Information" to the rescue. It's made life so
much easier.
> I do it all the time, and while tools like Firebug and Dreamweaver's tag
> inspector help they're not infallible -- add a bit of scripting or (in the
> case of DW) a CMS and they quickly become inaccurate.
We don't have too much scripting ATM, so it's been OK up to now.
> That said, I'm not at all certain that eliminating the cascade would help
> any -- having to define font styles for every element, for example, would
> get very tedious very quickly and send a whole lot of extraneous crap down
> the pipe. One could potentially mitigate the problem with an entity/macro
> system, but that would only cut down on the characters sent, not on the
> complexity of adding the appropriate styles to all possible elements --
> particularly for sites where you'll be displaying user-generated content.
>
> So I'm not sure there's a good solution to the problem -- though I'd love to
> see someone try to find one.
Maybe I've been too close to CSS for too long, but conceptually it's a
very elegant solution. The implementation might suck sometimes, but
that keeps life interesting.
> His other point is that CSS is 'broken' because each browser interprets it
> differently. That's a yes-and-no situation. It's true that the CSS standard
> isn't precise enough to expect exact agreement from all user agents -- but
> that's a necessity given the number of environments in which CSS formatting
> will be displayed.
>
> OTOH, even if CSS 2.1 *were* so precise and bug-free that it could be
> implemented perfectly by every UA, there's no way to force UAs to actually
> implement it completely and without error. And it's browser bugs and
> incomplete support that are the root of 99% of the trouble -- and those have
> squat to do with 'standards organisations'.
Indeed.
Tim