[css-values-4] Recursive math functions are inconsistently specified

[AtkinsSJ]

The spec about math functions is a bit inconsistent, at least from what I can tell. Each math function is described as allowing a calculation for its parameters, but that links specifically to the calc() definition, whereas the examples show that other math functions are allowed there. The syntax description does not itself allow any function (including calc()) to appear as a parameter to another math function.

So the issues here are:

• The syntax should allow for math functions as part of <calc-value>.
• The places that currently link to calculation should link elsewhere. The usage of calculation includes literal numbers or values, and math functions, as well as calc-like expressions. (Which generally corresponds to <calc-value>.)

[AtkinsSJ]

This is also a problem with the previous, simpler definition from CSS-VALUES-3. Strictly following the given syntax means that calc(3 + calc(4 * 2)) is invalid, when it should be valid.

[tabatkins]

Each math function is described as allowing a calculation for its parameters, but that links specifically to the calc() definition, whereas the examples show that other math functions are allowed there.

Specifically, they link to the "calculation" definition, which happens to be placed in the section defining calc(); they're not linking to calc() itself. The definition is not calc()-specific; it describes the argument to calc() (and the other math functions): "a sequence of values interspersed with operators, and possibly grouped by parentheses (matching the <calc-sum> grammar)"

The syntax description does not itself allow any function (including calc()) to appear as a parameter to another math function.

They do. All the grammar definitions eventually bottom out in <calc-value>, which is <number>/<dimension>/etc. Math functions match those productions based on what they contain, as specified in Type Checking.

[AtkinsSJ]

Thank you! And sorry about that. 😅