g++ seems to do a different name mangling than dmc, so this only works on windows.

[file: test1.cpp]
#include <stdio.h>
void foo() { printf("Hi from C++\n"); }

[file: test2.d]
extern (C++) void foo();
void main() { foo(); }

> dmc -c test1.cpp
> dmd test2.d test1.obj

I guess it's good to have a shared backend.

-----------------------
Carlos Santander Bernal

Carlos Santander B. wrote:
> g++ seems to do a different name mangling than dmc, so this only works on
> windows.
> 
> [file: test1.cpp]
> #include <stdio.h>
> void foo() { printf("Hi from C++\n"); }
> 
> [file: test2.d]
> extern (C++) void foo();
> void main() { foo(); }
> 
> 
>>dmc -c test1.cpp
>>dmd test2.d test1.obj
> 
> 
> I guess it's good to have a shared backend.
> 
> -----------------------
> Carlos Santander Bernal
> 
> 
> 
So there is some link compatibility with C++ on windows. Wonder if
Walter is aware of this. I rewrote test1.cpp to use iostream and it still worked. I will continue to play around with this undocumented feature.

Since gdc uses the same backend as g++, perhaps it can be done on Unix as well.

If that works and once I've discovered how much of C++ I can link my D projects with, I go back to that FOX toolkit port I shelved. Being able to link against C++ object should make it easier, rather than rewriting the whole dang thing in D(the reason it's been shelved).

--
Robert Jones
robertjones21@HotPOP.com

"Robert Jones" <robertjones21@HotPOP.com> wrote in message news:cd0vg3$ik4$1@digitaldaemon.com...
> So there is some link compatibility with C++ on windows. Wonder if Walter is aware of this. I rewrote test1.cpp to use iostream and it still worked. I will continue to play around with this undocumented
feature.

This is just C++ functions, not C++ objects/methods, right?

-david

David Barrett wrote:
> "Robert Jones" <robertjones21@HotPOP.com> wrote in message
> news:cd0vg3$ik4$1@digitaldaemon.com...
> 
>>So there is some link compatibility with C++ on windows. Wonder if
>>Walter is aware of this. I rewrote test1.cpp to use iostream and it
>>still worked. I will continue to play around with this undocumented
> 
> feature.
> 
> This is just C++ functions, not C++ objects/methods, right?
> 
> -david
> 
> 
That is what I intend to find out.

>> 
>> This is just C++ functions, not C++ objects/methods, right?
>> 

As far as I can say C++ struct and functions work, but as soon as there is any class around - even if it's not accessed - thing screw up.

system: linux / gcc 3.4.

Thoas

Thomas Kuehne wrote:

>>>This is just C++ functions, not C++ objects/methods, right?
>>>
> 
> 
> As far as I can say C++ struct and functions work, but as soon as there is
> any class around - even if it's not accessed - thing screw up. 
> 
> system: linux / gcc 3.4.

Did you use gdc/gcc to run your test?

> Thoas

Robert Jones wrote:

>>>>This is just C++ functions, not C++ objects/methods, right?
>> 
>> As far as I can say C++ struct and functions work, but as soon as there is any class around - even if it's not accessed - thing screw up.
>> 
>> system: linux / gcc 3.4.
> 
> Did you use gdc/gcc to run your test?

No, dmd 0.95

a primitive example is: (note that only "add" has C linkage but "multi" C++)

----a.cpp----
typedef struct MyStruct{
        int     i;
        char    c;
};

void multi(MyStruct* arg){
        arg->i = arg->i * 3;
}

extern "C" {
        void add(MyStruct* arg);
}

void add(MyStruct* arg){
        arg->c = arg->c+1;
        multi(arg);
}


----b.d----
extern(C){
        struct MyStruct{
                int i;
                char c;
        }
        void add(MyStruct* );
}

int main(){
        MyStruct m; m.i=1; m.c='A';

        printf("%i %c\n", m.i, m.c);
        add(&m);
        printf("%i %c\n", m.i, m.c);

        return 0;
}