1. Read it as:

Call expressions where one of the following conditions is true are not core constant expressions ([expr.const]):

• id >= num_args_
• or the type of the corresponding format argument (after conversion to basic_format_arg) is not one of the types in Ts...

2. It has no Effects, because it has no effects. The only effect is to be a core constant expression, or not. That's so that you get an error if the conditions are not met when the consteval function checks the format string. It has no Returns because its return type is void.

3. No, it doesn't mean that. It means "expressions that are calls of this function", i.e. invocations of this function.

1. Read it as:

Call expressions where one of the following conditions is true are not core constant expressions ([expr.const]):

• id >= num_args_
• or the type of the corresponding format argument (after conversion to basic_format_arg) is not one of the types in Ts...

Yeah thanks, it took me a while. It's pretty late here and the run-on sentence just really messed with my head.

It has no Effects, because it has no effects. The only effect is to be a core constant expression, or not. That's so that you get an error if the conditions are not met when the consteval function checks the format string. It has no Returns because its return type is void.

I've read P2757R3 and it sort of makes sense, but it's quite the surprising function and I feel like this deserves a note which explains the intent at least.

No, it doesn't mean that. It means "expressions that are calls of this function", i.e. invocations of this function.

I would have much preferred the wording "Invocations of this functions" over "Call expressions". It wasn't obvious to me that the very general statement about expressions is meant to apply to calls to this particular function. Maybe that's just me, but it stumped me. I can't think of having seen "Call expressions" used in such a way anywhere else in the standard either.

What do you think about the following wording?

An invocation of this function is a core constant expression only if

• id < num_args_ is true and
• the type of the corresponding format argument (after conversion to basic_format_args<Context>) is one the types in Ts....

The old wording was unnecessarily complex in terms of grammar. This grammatical sandwiched wasn't reader-friendly and I'm sure I'm not the last person to be stumped by it.

Furthermore, the negation was counter-productive. It defined that it's "note a constant expression if the type is not one of ...". Such double negations are confusing to the reader and it is much easier to express when a call is a constant expression.

I guess a note would also be nice, along the lines of:

This function has no effect. It is meant to verify during constant evaluation whether an argument typematches one of T....