Re: [CSS3, border-radius] and now is real fun.

Zack Weinberg wrote:
> Andrew Fedoniouk <news@terrainformatica.com> wrote:
>   
>> Slightly dashed and slightly dotted sounds like "slightly pregnant"
>> for me.
>> Or does CSS allow such state? If "yes" then how. If "no" then also
>> how. We cannot just be silent about it.
>>     
>
> Sure we can.  It's perfectly fine for the rendering of style
> transitions around curved corners to be unspecified in Level 3.
>
>   
>> If it is undefined then we need to say so clearly and to provide any 
>> reasonable fall-back. I personally do not see any other options
>> other than treating corner curves as having solid color always.
>> But that's me, AFAIR someone here said that they have a solution and
>> current rendering in Mozilla is just a bug. So asking.
>>     
>
> That was me.
>
> To be clear, when I said "current rendering in Mozilla is just a bug" I
> was thinking first of this case:
>
>   border-style: dotted;
>   border-radius: <nonzero>
>
> where the dots obviously should continue around the corner curve.
> Mozilla currently draws that with solid corners, and that's a bug.
> (Specifically, https://bugzilla.mozilla.org/show_bug.cgi?id=382721 .)
>
> Mozilla has no current plans to implement anything but sharp
> transitions in the middle of the corner. I would argue that that is
> clearly preferable to "always draw the corner as solid", no matter what
> the adjoining styles are.
>
> For the specific case of dotted transitioning to dashed I do not see
> what would be wrong with having the dots stretch into dashes along the
> curve; the pregnancy analogy makes no sense.
>   
dotted to dashed transition is probably not that difficult  to imagine. 
If to close our eyes on
complexity of rasterisation in this case.

I am speaking about cases like transition from dotted to double line as 
Brad mentioned.
Strictly speaking it is even physically impossible on pixel grids of 
finite size.
Or at least it should be some unknown yet non-trivial morphing rules.

--
Andrew Fedoniouk.
http://terrainformatica.com

Received on Tuesday, 25 August 2009 23:17:41 UTC