同样的可变参数函数, 为什么对float类型就不行呢?

[lightnut]

某哥们的原贴.
http://topic.csdn.net/u/20090220/11/951b405d-802c-4266-a0e4-da4f2b0991e9.html
我试图去解决上贴的问题, 失败了:(.

我用模板重写了上贴的可变参数函数, 发现对int, double类型没有问题, 惟独float的结果错误.
不只为何? (编译环境:VC2005)

#include <iostream> 
#include <cstdarg> 

template <class T>
T sum(int n,T b,...) 
{ 
   va_list ap; 
   T temp;
   va_start(ap, b); 
   T sum1=b;
   for(int i=0;i<n-1;i++) 
   { 
      temp=va_arg(ap,T); 
      sum1 += temp; 
   } 
   va_end(ap); 
   return(sum1); 
}

int main() 
{ 
   int intA = 2;
   int intB = 3;    
   std::cout <<intA<<"+"<<intB<<"="<<sum(2, intA, intB)<<'\n';

   double doubleA = 2.2;
   double doubleB = 3.3; 
   std::cout <<doubleA<<"+"<<doubleB<<"="<<sum(2, doubleA, doubleB)<<'\n';

   // 下面输出的结果是错误的！！ why? 
   float floatA = 2.1f;
   float floatB = 3.1f; 
   std::cout <<floatA<<"+"<<floatB<<"="<<sum(2, floatA, floatB)<<'\n';

   return 0;
}

[lightnut]

怎么忘记加分了? 200慢慢加........

[fox000002]

看看 g++ 的提示

test.cpp: In function ‘T sum(int, T, ...) [with T = float]’:
test.cpp:33:   instantiated from here
test.cpp:13: warning: ‘float’ is promoted to ‘double’ when passed through ‘...’
test.cpp:13: warning: (so you should pass ‘double’ not ‘float’ to ‘va_arg’)
test.cpp:13: note: if this code is reached, the program will abort

[lightnut]

[Quote=引用 2 楼 fox000002 的回复:]
看看 g++ 的提示



C/C++ codetest.cpp: In function ‘T sum(int, T, ...) [with T = float]’:
test.cpp:33: instantiated from here
test.cpp:13: warning: ‘float’ is promoted to ‘double’ when passed through ‘...’
test.cpp:13: warning: (so you should pass ‘double’ not ‘float’ to ‘va_arg’)
test.cpp:13: note: if this code is reached, the program will abort
[/Quote]

原来是这样!? 我将Warning开到最高, VC也不给提示.

[lightnut]

自己找到答案:

http://c-faq-chn.sourceforge.net/ccfaq/node266.html

``参数默认晋级" 规则适用于在可变参数中的可变动部分: 参数类型为 float 的总是晋级 (扩展) 到 double, char 和 short int 晋级到 int。所以 va_arg(arpg, float) 是错误的用法。应该总是用 va_arg(arpg, double)。 同理, 要用 va_arg(argp, int) 来取得原来类型是 char, short 或 int 的参数。基于相同理由, 传给 va_start() 的最后一个 ``固定" 参数项的类型不会被晋级。