W3C home > Mailing lists > Public > www-style@w3.org > May 2004

Re: CSS Gradients

From: Andrew Fedoniouk <news@terrainformatica.com>
Date: Thu, 13 May 2004 23:36:34 -0700
Message-ID: <002e01c4397d$ccfb4110$0301a8c0@ATHLON>
To: <ernestcline@mindspring.com>, "W3C CSS List" <www-style@w3.org>

> Rectangular gradients make little sense for borders and absolutely
> no sense for outlines.

What is the "outline" here? Isn't it a border?

Anyway....
If you wish to have gradients on borders it is well better to use something
a la:

border-color-gradient=gradient(color,color,color,color)

as this will also apply to rounded corners (If we will have them...)

And about this famous O(n^2)...

To build a linear gradient bitmap you need  three (four)  integer increments
+
three (four)  integer comparisons per point. (My guess)
I don't think that it is a big deal for any device these days to do such
math.

Andrew Fedoniouk.
http://terrainformatica.com

> > I would suggest to use four separate corner colors instead of two
> > colors and direction specification.
> >
> > background-gradient=gradient(color1,color2,color3,color4);
> >
> > Don't need to specify rotation or angle in this case. Already there.
> > And it's easy.
>
> Easy to specify, but not easy to compute.  It's not difficult to compute
> either in my opinion.  When dealing with graphics anything that for
> an n by n picture is O(n^2) is average, less than that is easy, and
> worse than that is hard in my opinion.  But my objection to rectangular
> gradients has little to do with their computation complexity (which is
> only worse by a scaling factor than a freely rotated linear gradient)
> but primarily with their lack of general applicability.
>
> Rectangular gradients make little sense for borders and absolutely
> no sense for outlines.  If gradients should be added to CSS at all,
> it should be done in a way that mixes well with all color properties,
> not just some of them.  I don't want to special case the foreground
> and background by having rectangular gradients apply to them
> but not to the border and outline.
>
> If you can point out a sensible way to apply rectangular gradients
> to borders and outlines and still get the same effect as this proposal
> does, then I'd be for it, but I just don't see it.
>
>
Received on Friday, 14 May 2004 02:37:20 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 27 April 2009 13:54:30 GMT