[css-values-5] Should progress functions have a calculation tree as their internal representation?

[cdoublev]

Like math functions. This would help to clarify:

• how to determine their type when they are used in a calculation
• whether the result of progress() can be unwrapped during the simplification of a calculation tree
• whether their resolved value should be clamped to the range allowed in the target context
• whether calc(progress(...)) should serialize as is or with progress(...) as a component of a declared value

[tabatkins]

I've now (in 5afd5de) marked progress() as a math function, so it gets caught by the general terminology about internal representation.

The other two are not math functions, just ordinary functions that resolve to a number, and evaluate their own calculation arguments in a special context. I've added a bit of text to Values 4 (in https://drafts.csswg.org/css-values-4/#calc-type-checking) about how contexts get determined, which hopefully makes this clearer.

[cdoublev]

Thank you, this is useful.

About clarifying how to determine the type of media-progress(), container-progress(), and also tree-counting functions... the specs are actually already clear: they are <number>s.

Anything else is a terminal value, whose type is determined based on its CSS type.

For what is worth, I prefer to not consider them as <number>s, but as resolving to a <number>.

Anyway, I am more interested in the observable output. For exemple, whether calc(media-progress(...)) should serialize with or without calc(), and whether media-progress() matching <integer> should be clamped at computed-value time.