I don't want to refer to a grammar term, decree that it "shall be empty", and then rely on an "equivalent" phrasing somewhere to justify that it nonetheless might not actually consist of zero tokens.

Maybe we should say "the parameter list shall be empty" or so, and have dcl.fct clearly define "parameter list". Maybe "parameter-type-list" is a term which would fit here.

I think we should define the term "parameter list", which is a list that comprises each parameter introduced by the corresponding declaration in parameter-declaration-clause, In other words, the sequence of parameters. "parameter-type-list" is the sequence of the (adjusted)types of the parameters. The former is more suitable here since we are only concerned about whether the destructor has parameters. Instead, the latter is defined as producing types.

parameter-type-list also represents the presence or absence of the va_args ellipsis, I think.

Anyway, a parameter list can also represent the ellipsis(syntactically, it's , ...). It might make the intent clear that we separate the parameters from parameters' types. The definition of "parameter list" might be

A parameter list consists of each parameter introduced by parameter-declaration and ... that is part of parameter-declaration-clause(if any).

And we have said that

A parameter list consisting of a single unnamed parameter of non-dependent type void is equivalent to an empty parameter list.

So, we rule that the parameter list of a prospective destructor shall be empty, which clearly means

• either its parameter-declaration-clause consists of zero token, or
• its parameter-declaration-clause is just void.

I think.