[cairo] Re: cairo c++ bindings

[Bill Spitzak]

Benoit Perrot wrote:

> It seems to me that deriving from the C struct is highly dangerous:
> what happen if, for a reason or another the compiler decides to add
> informations before the "relevant" data block?
> Like in:
> 
> class Matrix : public cairo_matrix_t
> {
>   void init_identity()
>   {
>     cairo_matrix_init_identity(this);
>   }
> };

This is actually ok as the call will actually conversion-cast the 
Matrix* to a cairo_matrix_t*. This means it will move the pointer to 
after any possible vtab that Matrix has. If it didn't do this, it would 
not be possible to add virtual functions to a subclass of a 
class/structure that does not have any. It would also be impossible to 
do multiple inheritance.

What is a worry is that C++ will define the structure slightly 
differently than C. But so much code relies on this that I think as long 
as the cairo.h header file is in "extern "C" {...}" you can assumme it 
will work.

Also I certainly do want any C++ wrapper to be inline, at least for all 
methods that translate to a single cairo function call. Doubling the 
procedure call overhead is unacceptable and will prevent any serious 
users from using the C++ wrapper.