Work on SERVER-81489 has brought to light a syntactic problem in the way IDL specifies the names of C++ functions that it imports and uses.

It tries to determine whether a specification is referring to a member function or nonmember function.
This is not generally possible to do. An identifier can mean either thing, even if it has :: scope operators embedded, as these could still be referring to a base class member function, like:
"myns::func" could refer to an intent to specify a member function in the base class myns, invoked as: item.myns::func(). So what it does is consider something to be a nonmember function only if it has a LEADING scope operator, but this would anchor the function in global scope. It's unclear syntax.

I think we should look into the idea of extending IDL syntax so that a leading . dot will indicate member function unambiguously.
EVERYTHING else would be a nonmember function. I think we can probably sweep the codebase and make this change in a reasonable timeframe.