开发者

complex C++ template syntax

After joining SO, I am frequently seeing this kind of syntax whenever I open topics discussing templates. I tried searching on google, but in vain.

template<typename T>
char (&f(T[1]))[1]; //what is it? what is the use of '[]' brackets and the integer in it?

template<typename T>
char (&f(...))[2]; //not this either

int main() { char c[sizeof(f<void()>(0)) == 2]; } // and this?

From here : SFINAE with invalid function-type or array-type parameters?

Please explain the 3 lines where I have put comments. I particularly want to understand the syntax. Can w开发者_开发知识库e use such syntax in templates only?


The following two are equivalent

// 1
template<typename T>
char (&f(...))[2]; 

// 2
typedef char rettype[2];
template<typename T>
rettype &f(...);

You may have have seen that pattern before with function pointers

char (*XXX)();

Now just replace the () with [N] to create an array instead of a function part, and replace * by & to create a reference instead of a pointer, and replace the XXX by a function declarator. Then you get a function that returns a reference to an array of size N.


You may want to look into man signal, which contains a similarly typed function declaration. If you take out the inner declarator that actually declares the function, you get the same pattern

void (* signal(int sig, void (*func)(int)) )(int);
//      ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ take out that 

It will return a pointer to a function that takes an int and returns void, as described in that manpage.


The following is just a way to yield a compiler error if some condition is not satisfied. If the test foo == 2 turns out to be false, the array created is of zero size, which is illegal in C++, and will earn a compile time error. If it evaluates to true, nothing happens except the array being declared.

char c[some silly condition here];
0

上一篇:

下一篇:

精彩评论

暂无评论...
验证码 换一张
取 消

最新问答

问答排行榜