Re: Integration with C

[Andrew Haley <aph at redhat dot com>]

Geert Bevin writes:
 > 
 > thanks for the answers.
 > 
 > > Well, it depends on what exactly you mean.  I suppose that you want to
 > > call Java from C, and in turn you want Java to be able to call back
 > > into C.
 > 
 > What I want is a backend java library which can perform long operations 
 > in a background thread. A C/C++ GUI can link against this lib and 
 > register a number of callback hooks (as function pointers). Then a long 
 > operation is started in a seperate thread and the java library goes over 
 > the registered hooks (listeners) to report progress, messages, errors 
 > and finish. When the process is finished, the GUI is able to unregister 
 > the hooks too.
 > 
 > This would make is possible to write a reponsive GUI, that remains 
 > usable by a user while a long operation is being executed in the background.
 > 
 > > Define a Java Class thusly:
 > > 
 > > 
 > > class Callback
 > > {
 > >   RawData method;
 > >   public static native int foo (int bar, int baz);
 > > }
 > > 
 > > and the C++ thusly:
 > > 
 > > typedef int (*fptr)(int, int):
 > > jint foo (jint bar, jint baz)
 > > {
 > >   fptr f = (fptr)method;
 > >   (*f)(bar, baz);
 > > }
 > 
 > I see how this would work, but my main question is how do you get the 
 > function point in the method variable. Here you only use it, you never 
 > provide it. Since the interface needs to be declared in java as being 
 > native, there's no way to write a method that takes a function pointer 
 > argument, is there?

I'm totally mystified by where you think the problem might be.
Declare a Java native method that takes a RawData arg, and cast a
function pointer to RawData.  It's not legal C++, but it will work on
many architectures.

 > Would it be possible to derive the Callback class in C++ alone and
 > add a setCallback(fptr callback) method?

Sure, why not?  You can write all of a Java class's implementation in
C++ if you wish.

 > It would even be better if this would be possible in the Callback
 > class inself, but I suppose that's not an option.
 > 
 > > But why bother?  It might be easier just to keep a cache of function
 > > pointers somewhere and pass around indexes instead.
 > 
 > So you'd store a list of function pointers in a C++ class and work with 
 > integer position identifiers instead. That feels a bit 'dirty' to me ;-) 

Perhaps, but it is legal C++, and casting function pointers to data
pointers isn't.  

 > I like to work was as much typed information as possible.

You lose no typed information by doing this.  Declare the array of
function pointers with the correct type.

Andrew.