Re: [gradients] basics

On Dec 8, 2009, at 1:28 PM, "Tab Atkins Jr." <jackalmage@gmail.com> wrote:

> [Not quoting because I don't have particular points to respond to.]
> 
> All right, Brad, you're right.  And wrong.  ^_^  Your reasoning is
> wrong - you're either thinking about something incorrectly,

Unlikely.   ;)

> [..]
> But that's no good; that means that even the default 'auto' value of
> background-size is incoherent when your background is a gradient.  We
> need to fix this.  This is where your idea comes in; saying that the
> gradient *is* finite in size and has edges equal to the 'box' used
> when generating them is the most natural way to deal with this.  This
> is intuitive and makes gradients work nicely with all the background
> properties.

Good. It was so intuitive to me that I never considered that it would be any other way (pretty much any image is going to need some sort of logical box to draw it into). But we are at least in agreement now.

> Now we come to background-repeat.  You're correct that this is the
> right place to make a gradient extend out of that box, because this
> property actually transforms finite-size images *into* infinite-size
> ones.  An image with background-repeat:repeat goes on forever in both
> dimensions, and no longer has edges, just like gradients.  So adding
> another value to this property, 'extend', which removes the visual
> clipping from naturally-infinite images is good and proper.  Just like
> a repeated background, the 'size' of the image is still the same as it
> would be without the repeating.

Exactly.

> So, this all sound sane?  I suspect it does, since I'm just explaining
> back to you what you already suggested.  

Yes. Much, much more sane to me.

> If so, then I'll change
> Images to explain this, and B&B'll need the new value added to
> background-repeat.  No other changes should be necessary.

I'm not sure if B&B needs to be changed (or held up), or if this can just be a gradients-dependant addendum that lives in the gradients spec.