[css3-2d-transforms] DOM interfaces, especially CSSTransformValue

[Øyvind Stenhaug]

1)
Section 7 of the CSS 2D Transforms draft specifies a number of DOM  
interfaces, among them CSSTransformValue which inherits from CSSValueList.  
However, according to an earlier message from the WG [1], interfaces  
inheriting from and related to CSSValue are "classified as obsolete", and  
"[i]mplementors are strongly urged not to implement them". This would seem  
to apply to CSSTransformValue and probably also CSSMatrix as written.

2)
It seems inconsistent/confusing that "getComputedStyle([...]).transform"  
would return a CSSValue, since it looks just like the DOM attributes in  
CSS2Properties which are all of type DOMString and equivalent to calling  
getPropertyValue.

[1] http://lists.w3.org/Archives/Public/www-style/2003Oct/0347.html

--
Øyvind Stenhaug
Core Norway, Opera Software ASA

[Dean Jackson-7]

On 10/09/2009, at 2:56 AM, Øyvind Stenhaug wrote:

> 1)
> Section 7 of the CSS 2D Transforms draft specifies a number of DOM  
> interfaces, among them CSSTransformValue which inherits from  
> CSSValueList. However, according to an earlier message from the WG  
> [1], interfaces inheriting from and related to CSSValue are  
> "classified as obsolete", and "[i]mplementors are strongly urged not  
> to implement them". This would seem to apply to CSSTransformValue  
> and probably also CSSMatrix as written.

Yeah. Do you have a suggestion on how to fix this? What is the best  
alternative?

I'll note that the earlier message is from 2003, and says "We don't  
have a Working Draft yet" :)

> 2)
> It seems inconsistent/confusing that "getComputedStyle
> ([...]).transform" would return a CSSValue, since it looks just like  
> the DOM attributes in CSS2Properties which are all of type DOMString  
> and equivalent to calling getPropertyValue.

What would you prefer?

Dean

[Øyvind Stenhaug]

On Sat, 31 Oct 2009 01:22:42 +0100, Dean Jackson <dino@...> wrote:

>
> On 10/09/2009, at 2:56 AM, Øyvind Stenhaug wrote:
>
>> 1)
>> Section 7 of the CSS 2D Transforms draft specifies a number of DOM  
>> interfaces, among them CSSTransformValue which inherits from  
>> CSSValueList. However, according to an earlier message from the WG [1],  
>> interfaces inheriting from and related to CSSValue are "classified as  
>> obsolete", and "[i]mplementors are strongly urged not to implement  
>> them". This would seem to apply to CSSTransformValue and probably also  
>> CSSMatrix as written.
>
> Yeah. Do you have a suggestion on how to fix this? What is the best  
> alternative?

No suggestions, sorry... unsure of the use cases of these interfaces and  
not very familiar with CSSOM things.

>> 2)
>> It seems inconsistent/confusing that "getComputedStyle
>> ([...]).transform" would return a CSSValue, since it looks just like  
>> the DOM attributes in CSS2Properties which are all of type DOMString  
>> and equivalent to calling getPropertyValue.
>
> What would you prefer?

That it return a string just like getPropertyValue('transform'). What  
format exactly I'm not sure (that isn't clear for other properties  
either). Looks like WebKit and Gecko return a stringified matrix function  
for transformed elements and "none" otherwise.

If the CSSValue interface were not obsoleted, it looks like it would be  
expected to be accessible as getPropertyCSSValue('transform').

--
Øyvind Stenhaug
Core Norway, Opera Software ASA

[Sylvain Galineau]

> From: www-style-request@... [mailto:www-style-request@...] On
> Behalf Of Dean Jackson

> On 10/09/2009, at 2:56 AM, Øyvind Stenhaug wrote:
>
> > 1)
> > Section 7 of the CSS 2D Transforms draft specifies a number of DOM
> > interfaces, among them CSSTransformValue which inherits from
> > CSSValueList. However, according to an earlier message from the WG
> > [1], interfaces inheriting from and related to CSSValue are
> > "classified as obsolete", and "[i]mplementors are strongly urged not
> > to implement them". This would seem to apply to CSSTransformValue
> > and probably also CSSMatrix as written.
>
> Yeah. Do you have a suggestion on how to fix this? What is the best
> alternative?

See my last mail on this topic [1]. Once we have agreement on the obsolete
interfaces, we can request DOM L2 Style to include an informative note similar
to the red box at the top of CSS2 [2] noting the interfaces considered obsolete.

[1] http://lists.w3.org/Archives/Public/www-style/2009Nov/0259.html
[2] http://www.w3.org/TR/2008/REC-CSS2-20080411/

[Dean Jackson-7]

On 19/11/2009, at 9:58 AM, Sylvain Galineau wrote:

>> From: www-style-request@... [mailto:www-style-request@...] On
>> Behalf Of Dean Jackson
>
>
>> On 10/09/2009, at 2:56 AM, Øyvind Stenhaug wrote:
>>
>>> 1)
>>> Section 7 of the CSS 2D Transforms draft specifies a number of DOM
>>> interfaces, among them CSSTransformValue which inherits from
>>> CSSValueList. However, according to an earlier message from the WG
>>> [1], interfaces inheriting from and related to CSSValue are
>>> "classified as obsolete", and "[i]mplementors are strongly urged not
>>> to implement them". This would seem to apply to CSSTransformValue
>>> and probably also CSSMatrix as written.
>>
>> Yeah. Do you have a suggestion on how to fix this? What is the best
>> alternative?
>
> See my last mail on this topic [1]. Once we have agreement on the  
> obsolete
> interfaces, we can request DOM L2 Style to include an informative  
> note similar
> to the red box at the top of CSS2 [2] noting the interfaces  
> considered obsolete.

For now I've removed the interfaces that inherited from CSSValue.

Dean

> [1] http://lists.w3.org/Archives/Public/www-style/2009Nov/0259.html
> [2] http://www.w3.org/TR/2008/REC-CSS2-20080411/
>