Re: Apple's Proposal for CSS Animation

On Nov 8, 2007, at 10:29 AM, Brad Kemper wrote:

> David, I think you may wish to state in your spec that if non- 
> animatable CSS properties are included (as they especially might be  
> with the keyword of "all") that the transition still takes place on  
> the animatable properties, i.e. it does not cause the entire  
> transition from taking place due to invalidity. If there are no  
> valid animatable properties included then the whole transition  
> would be ignored.
>

Yeah, I agree.  This is actually what is implemented in WebKit right  
now.

>> Is "none" even a useful value if the initial value is "all"? The  
>> syntax is more elegant if transition-duration defaults to 0 and  
>> this property defaults to "all", but another option is to default  
>> this property to "none" and duration to something reasonable,  
>> e.g., 250ms. This would force an author to specify transition- 
>> property in the shorthand all the time though.
> "None" would still be useful as a handy and intuitive way to kill  
> the transition, I would think, either in JavaScript or if you  
> wanted to prevent the transition in certain elements that were part  
> of a class that received transitions. I agree with your elegance  
> argument. I would be more likely to want to assume "all" and  
> declare the duration, since durations will almost always be unique,  
> but "all" will be by far the most common for "transition-property".
>
> Although it does confuse the shorthand if there are two different  
> things that can be "none", doesn't it? For instance, with this:
>
> 		#this { transition: left none ease-in 2s; }
>
> "left" could be the name of a CSS property, or it could be a  
> transition-name, could it not?

At the moment transition-names have to be strings, so there's no  
ambiguity right now.  I kind of hate this though (as did Bjoern), but  
as you say, if they are both identifiers there is an ambiguity.  This  
could perhaps be resolved simply by stating that the first identifier  
encountered is always transition-property.  Perhaps there is  
precedent for how to resolve this issue with an existing CSS shorthand.