Re: box-shadow and border-image and dashes

On Feb 19, 2009, at 1:09 AM, Ph. Wittenbergh wrote:

>
> On Feb 19, 2009, at 1:14 AM, Brad Kemper wrote:
>
>> By the way, for regular (non-image) borders with border-styles that  
>> contain empty space (dashed, dotted, double, etc.), the current  
>> implementations are not consistent in how they handle box-shadow  
>> with background-clip:padding-box. WebKit continues the shadow in  
>> the gaps (e.g. between the dashes), and FireFox/MineField does not  
>> (not the version I'm looking at anyway, so maybe that is actually a  
>> bug).
>
> Gecko looks a bit odd, to be honest (and WebKits dotted border is  
> ugly, but that is unrelated).

Yes, I've compared dotted line before on this list. I've always  
thought that dots should be as round as possible. Otherwise they are  
square. IE7 was the best I tested with, because it distributed space  
between dots so that the corners were also dots and the entire border  
was symmetrical. Firefox's corners are not too pretty.

>
> Which behaviour do you expect ?

My preference (and initial expectation) would be to have the box- 
shadow follow the contours of silhouette created by the union of the  
shapes in the border (the dashes, dots, etc., but not the spaces  
between them) and the background shape (i.e. the border-box or padding- 
box).

I'm told that goes beyond "box-shadow" and gets into general "shadow"  
territory though, for which there is no spec currently (unless you  
include SVG filters, which I haven't tried yet). So my second choice  
would be to do the WebKit way and have the shadow continue behind any  
gaps in the border. Having no shadow in those gaps is completely  
unexpected to me.

It would also be useful to have an option for a corresponding shadow  
generated for the elements of the background (images and colors with  
alpha) based on their transparency/opacity, similar to what is being  
proposed for image-border. If that happened, it should be switchable  
via a keyword, as it would not always be appropriate.

Received on Thursday, 19 February 2009 16:09:32 UTC