Dick Hendrickson writes:
> I think it is non-standard. section 12.4.1.5, restrictions on dummy
> arguments not present says
> "an optional dummy argument that is not present...
> shall not be referenced or defined."
> To me, that clearly prevents it from being used as an I/O
> inquiry spec if it is not present.
Agree. That's the most fundamental of the restrictions. The various
restrictions on use as an actual argument in some cases are the
complicated ones. They arise because use as an actual argument
doesn't necessarily count as a reference or definition, but you still
need to prevent many cases of it. Just because those complicated
cases are there, don't forget the simple one. Anything that
involves evaluation is going to be a reference - that includes
the cited I/O specifiers.
P.S. Of course, both compilers in question are conforming because
this restriction isn't a constraint.
--
Richard Maine
[log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|