--On Sunday, April 13, 2008 5:45 PM -0700 Dan Fandrich
<dan_at_coneharvesters.com> wrote:

> On Sun, Apr 13, 2008 at 01:42:08PM -0400, Nathan E. Moore wrote:
>> Section 7.15.1.1 The va_arg macro (also available at
>> <http://www.opengroup.org/onlinepubs/009695399/basedefs/stdarg.h.html>)
>> ...
>> The parameter type shall be a type name specified such that the type of
>> a pointer to an
>> object that has the specified type can be obtained simply by postfixing
>> a * to type. If
>> there is no actual next argument, or if type is not compatible with the
>> type of the
>> actual next argument (as promoted according to the default argument
>> promotions), the
>> behavior is undefined, except for the following cases:
> ^^^^^^^^^^^^^^^^^^^^^
>> — one type is a signed integer type, the other type is the
>> corresponding unsigned integer
>> type, and the value is representable in both types;
>> — one type is pointer to void and the other is a pointer to a
>> character type.
>>
>> I guess the way I read it says that you do get argument promotion. I
>> would assume the
>> promotion would be done prior to pushing the arguments onto the stack.
>> But I have been
>> wrong before and likely will be again.
>
> There is now way the compiler would know the expected type at the time of
> calling in order to do the promotion. The behaviour in this case is, as
> documented, undefined.
>
>>>> Dan

Indeed there is no way for the compiler to know that information short of
solving the
halting problem, but that is not what I meant. I meant that floats are all
promoted to
doubles, smaller integral types to long/unsigned long, before the function
call is made.
This interpretation matches what the specs for printf say about the length
modifiers (see
<http://www.opengroup.org/onlinepubs/009695399/functions/fprintf.html>)

Nathan Moore
Received on 2008-04-14