从R调用C ++函数时出错-库加载，但函数不在表中

Question

Here is the C code:

// a.cpp
void double_me(int* x) {
  // takes a numeric input and doubles it
  *x = *x + *x;
}

I compile the code with

>R CMD SHLIB a.cpp

After that i run R and type following commands:

 dinfo <- dyn.load("a.so")
 .C("double_me",x=2)

This end with error: "double_me" is not on the list.

Now the question: dyn.load works fine, dinfo contains:

DLL name: a Filename: /Users/myusername/a.so Dynamic lookup: TRUE

But the function is not on the table:

is.loaded("double_me") [1] FALSE

How could it happen? This happens on macOS.

Answer 1

This is because you are using a.cpp ; C++ function names are "mangled" by the compiler. You can use your same code with the filename ac , compiling it just as you did, and get the following from R:

> dinfo <- dyn.load("a.so")
>  .C("double_me",x=2)
$x
[1] 2

Or, alternatively, you can add this line to the top of a.cpp :

extern "C" void double_me(int* x);

and get the following from R:

> dinfo <- dyn.load("a.so")
>  .C("double_me",x=2)
$x
[1] 2

Update: Why was the result above 2?

If you do not coerce the argument to the proper type, a copy may be made, such that your original value is not altered; if we coerce the value to be an integer as we should when using .C() , we get the expected result:

> dyn.load("a.so")
> .C("double_me", x = as.integer(2))
$x
[1] 4