[css-values-5] Accept <whole-value> as argument of first-valid()

[cdoublev]

Is first-valid() intentionally defined with <declaration-value> rather than <whole-value>? Should first-valid(toggle(red; green)) be invalid?

https://drafts.csswg.org/css-values-5/#typedef-first-valid

<first-valid()> = first-valid( <declaration-value> [ ';' <declaration-value> ]* )

[Loirooriol]

Accept <whole-value> as argument of first-valid()

I thought the same in 5fc63f5#r131685583

But e.g. consider color: first-valid(invalid(); green). This needs to be valid (and resolve to green), so we can't require invalid() to be a <whole-value>.

Should first-valid(toggle(red; green)) be invalid?

No, it's valid, since toggle(red; green) is a <declaration-value> (being a <whole-value> is not exclusive).

[cdoublev]

Ah yes, sorry, first-valid() arguments should be validated at computed value time:

If none of the arguments represent a valid value for the property, the property is invalid at computed-value time.

I am still not sure if toggle() nested in first-valid() is valid at computed value time (it is not allowed to be nested), or at parse time in other cases (eg. toggle(toggle(red); green) or mix(50%; toggle(red); green)).

Thanks for the clarification!

[Loirooriol]

I am still not sure if toggle() nested in first-valid() is valid at computed value time

toggle() is not preserved at computed-value time. https://drafts.csswg.org/css-values-5/#toggle-notation: "the computed value of toggle() is C_{n+1}"

it is not allowed to be nested

I think this means that you can't have a toggle() inside another toggle(). But it should be fine to have a toggle() inside a first-valid().

[tabatkins]

Yeah @Loirooriol is correct on all counts.