Re: outline-radius

(12/03/13 2:46), Boris Zbarsky wrote:
> On 3/12/12 12:35 PM, Paul Bakaus wrote:
>> and thus, I think it makes sense to optimize the
>> most common usecase.
> 
> Or give them a better tool than "outline", which has a very particular
> meaning and behavior, to solve the problem they're trying to solve. What
> problem _are_ they trying to solve?

I think the problem statement was relatively clear in the first message
from Paul:

(12/03/12 21:38), Paul Bakaus wrote:
> [snip]
> Particularly in web apps featuring sorting and/or drag & drop,
> outlines are god-sent as they don't modify layout – so in order to
> mark something as a drop target while dragging on top of it, outlines
> are invaluable to not cause any harm to the layout but highlight the
> target.

I think current properties like 'border-radius' and 'border-image'
probably won't address this requirement and perhaps someone might be
able to do a detailed analysis. (Would 'border-image' help if it's
painting order is changed? But is that technically possible?)

(12/03/13 2:46), Boris Zbarsky wrote:
> On 3/12/12 12:35 PM, Paul Bakaus wrote:
>> If single element, then curved outlines.
> 
> You mean if single box?  That's fine, but my point was that the spec
> needs to describe behavior in the multiple-box case too.

Details are fun, but I feel like every time when someone talks in this
way, it *feels like* someone is discouraging people from sending
proposals. (Though I do have great sympathy for the whole implementation
industry.)

(12/03/12 23:43), Brad Kemper wrote:
> Outlines should have outside outline radii equal to border-radius +
> outline-width + outline-offset.

One more question about this formula. Does this mean that a
'border-radius' corner with radius in one dimension being 0 shouldn't be
treated as both radii are 0 at that corner? Namely, is it

(0, r) + outline-width + outline-offset

or

(0, 0) + outline-width + outline-offset.

Note that the spec has this sentence:

  # If either length is zero, the corner is square, not rounded.

But I guess it doesn't imply the latter. Then, a relevant question is
whether we should start to allow negative border-radius (but treat it as
0 when it's negative in the corresponding radius
(outline/border/padding/content)).



Cheers,
Kenny

Received on Monday, 12 March 2012 19:49:19 UTC