[css-transitions-2] Spec text talks about **values** being transitionable & having particular "animation type", when it probably wants to talk about **properties**

[dholbert]

I have a clarity/correctness nit with this spec text that was recently added to css-transitions-2.

Spec link:
https://drafts.csswg.org/css-transitions-2/#transition-property-property

Specifically this block of text:
69f4ac7#diff-a9fb88c817f0afdb3cb3e08c28315c1174ef4e5d7db3e09c7dffe49a4a5a59dbR87-R91

I'll quote the text in question (emphasis added):

In CSS Transitions Level 2, property values are transitionable unless they have an animation type that is not animatable. Values with a discrete animation type are transitionable, and flip at 50% progress (p = 0.5).

ISSUE:
This spec text is written as if values themselves might be transitionable-or-not, and as if they might have particular value-specific "animation type" (of "not animatable" or "discrete").

This doesn't quite make sense -- the "animation type" concept is associated with a CSS Property rather than with its values, as defined at https://drafts.csswg.org/web-animations-1/#animation-type (particularly in the case of the "not animatable" animation type).

SUGGESTED FIX:
@fantasai I think this paragraph should be rewritten to talk about the properties being transitionable and having a particular "animation type", rather than the values -- does that make sense? Or if there's value-specific subtlety that we're trying to lean into here, we should perhaps make that more apparent.

(Note that the immediate next paragraph talks about "transitionable properties" so I think that is the meaning we're going for -- "transitionable" being associated with properties-as-a-whole rather than values.)

[dholbert]

This is particularly for the "not animatable" portion, which applies to a whole property.

"discrete" does have a bit more of a value-specific flavor; it's defined as a whole-property thing at https://drafts.csswg.org/web-animations-1/#animation-type , but the "by computed value" / "repeatable list" animation-types there do say "the property values combine as discrete" for some cases. I think that's what this new css-transitions-2 spec text is meaning to capture here when referencing "Values with a discrete animation type".

Here's some suggested possibly clearer/more-precise text for https://drafts.csswg.org/css-transitions-2/#transition-property-property , now distinguishing between "properties with a discrete animation type" vs. properties with other animation types which have values that "combine as discrete":

In CSS Transitions Level 2, properties are transitionable unless they have an animation type that is not animatable.
Properties with a discrete animation type are transitionable, and flip at 50% progress (p = 0.5).
More generally: for any transitionable property that has values that combine as discrete, these values transition by performing a flip at 50% progress (p = 0.5).

(Further bikeshedding welcome.)

[dholbert]

Alternate replacement-text -- slightly more concise/direct though maybe slightly less human-friendly:

In CSS Transitions Level 2, properties are transitionable unless they have an animation type that is not animatable.
If a transitionable property has a discrete animation type for all of its values, or has certain value combinations that combine as discrete, then these values transition by performing a flip at 50% progress (p = 0.5).

[dholbert]

(See also web-platform-tests/wpt#39871 which is about tests expecting values vs. properties-as-a-whole being transitionable)