Re: [CSS2.1] Background boundaries and attachment

Ian Hickson wrote:
> On Mon, 5 Jan 2004, fantasai wrote:
> 
>>>>You'd have to add an extra element solely for presentation.
>>>
>>>So how would you do the opposite (image fixed with the border of the
>>>element instead of scrolling with the content) without using extra
>>>elements?
>>
>>background-attachment: fixed;    /* fixed wrt viewport, which has the effect of being
>>                                     fixed wrt the border when the parent scrollbox
>>                                     isn't being scrolled -- which covers frames behavior*/
>>background-attachment: attached; /* hypothetical CSS3 property: fixed wrt border */
> 
> 
> background-attachment: fixed; /* fixed wrt viewport */
> background-attachment: scroll; /* fixed to element */
> background-attachment: content; /* fixed to content */
> 
> I don't understand why one is better or worse than the other. I seem to
> recall the decision to define it the way it is was more based on the
> weight of existing implementations at the time than preferences either way
> on the issue, since the two options are pretty much symmetric.

Having the 'scroll' keyword mean fixed-to-content would, I believe, be
more consistent with authors' expectations because fixed-to-content is
the behavior you get when you specify 'scroll' on the main canvas -- or
on any non-scrolling block. In these (most common) cases, 'scroll' _most
notably_ causes the text and the background scroll together in all
implementations. If you define it to mean fixed-to-element, then this
interpretation breaks on scrolling elements. /That/ is the problem.

(There's also the problem that the CSS3 Backgrounds Last Call does not
include a third value to allow either third possibility, but I suppose
one can assume for the sake of argument that it's not really last call.)

~fantasai

Received on Wednesday, 28 January 2004 22:30:30 UTC