Re: [css3-ui] using <image> instead of <uri> in cursor (Issue 44)

Agreed precisely with Dirk.

Let's stick with <url> in CSS3-UI for this per the methodology of only
accepting features (or expansion thereof) if there is demonstrated
interop.

We can expand to <image> in CSS4-UI.

> Unless there are two compatible browser implementations for <image> of course. In this case CSS3 UI should support <image> as specified in CSS3 Images with a normative reference to CSS3 Images. The tests should be written against CSS3-UI.

Exactly.

Tantek



On Wed, Nov 26, 2014 at 9:45 AM, Dirk Schulze <dschulze@adobe.com> wrote:
>
> On Nov 26, 2014, at 1:31 PM, Florian Rivoal <florian@rivoal.net> wrote:
>
>>>
>>> On 25 Nov 2014, at 20:37, Rick Byers <rbyers@chromium.org> wrote:
>>>
>>> On Tue, Nov 25, 2014 at 12:55 PM, Dirk Schulze <dschulze@adobe.com> wrote:
>>>
>>> On Nov 25, 2014, at 2:10 PM, Florian Rivoal <florian@rivoal.net> wrote:
>>>
>>> > Raised in issue 44: https://wiki.csswg.org/spec/css3-ui#issue44
>>> >
>>> > currently, the cursor property takes its images from a list of <uri>. It should use a list of <image> instead, to get access to the various functional notations defined for this value type. This is backward compatible as <uri> is a valid value of <image>.
>>>
>>> I agree for CSS UI level 4. As you said, <image> is backward compatible. However, I am not so sure if linear/radial gradients and all other <image> values are supported in all browsers. My suggestion is to change it in the next level.
>>>
>>> Right.  Blink (and I believe still WebKit) currently support only image-set.
>>
>> Or we can do like css-backgrounds-3 did:
>>
>> http://dev.w3.org/csswg/css-backgrounds-3/#value-types
>>
>> Which allows people who want to get ahead to do so by combining specs, while not making it a conformance breaking things to only support the <uri> subset of <image>.
>
> I do not object to this. However, it seems to move the testing efforts from CSS UI to CSS Images since the latter is overriding the definition of <image>. That makes me feel uncomfortable and I would still prefer to use <url> in CSS3 UI.
>
> Unless there are two compatible browser implementations for <image> of course. In this case CSS3 UI should support <image> as specified in CSS3 Images with a normative reference to CSS3 Images. The tests should be written agains CSS UI.
>
> Greetings,
> Dirk
>
>>
>>  - Florian
>

Received on Wednesday, 3 December 2014 16:55:34 UTC