- From: Ernest Cline <ernestcline@mindspring.com>
- Date: Thu, 29 May 2003 18:22:58 -0400
- To: www-style@w3.org
Ben Godfrey wrote:
> From Ernest Cline's message:
>
> > However, in order to do that you'd have to increase the complexity of
> > the CSS parser so that it can handle unknown keywords.
>
> The keywords would be of a different format such as %keyword% or
> ${keyword}. The increase in complexity would just be a symbol table
> would it not? If we enforced that the constants must be defined before
> they are referenced, we would still only need one pass, I think. The UA
> could then drop the table before parsing user styles.
At the cost of for many properties being unable to engage in
optimizations that they engage in now because they only have to support
a finite set of token values, while supporting constants would mean
that they cannot engage in such optimizations. Despite what some might
think, I'm not opposed to optimization nor an automatic supporter
of every neat feature. If adding constants let authors do things that
they otherwise could not, or provided a means of decreasing document or
stylesheet size, then I'd be all for it, as I agree that it is a neat
feature. However it doesn't do either of those things and it prevents
useful optimizations from taking place.
> > It increases the workload involved in interpreting CSS for what would
> > be at most a marginal gain in usability, with no gain in capability.
>
> I'm not sure how great the cost increase would be and it would be a
> neat feature.
The main problem is that in my opinion the cost that would be borne by
the many to support this feature far outweighs the benefit that a few
would gain.
Received on Thursday, 29 May 2003 18:23:47 UTC