Implement InfoMapper in subclass

[Rod B]

Hi,

Can InfoMapper be implemented as a subclass? Or is there some way to tell the builder that a typedef already exists and not to generate a stub for it?

For instance in my header file I have:

    /**
     * Connection Error callback type
     */
    using ConnectionErrorCallback = std::function<void(ConnectionError)>;

and I am trying to make that type available to me in Java by making this class:

@Platform(
        include = {
                "mavsdk/core/include/mavsdk/mavsdk.h"
        }
)
@Name("mavsdk::Mavsdk::ConnectionErrorCallback")
public abstract class ConnectionErrorCallback extends FunctionPointer {
    static { Loader.load(); }

    /** Cast from a raw pointer (rarely needed). */
    public ConnectionErrorCallback(Pointer p) { super(p); }

    /** Allocate a new C++‐side std::function<void(ConnectionError)> stub. */
    public ConnectionErrorCallback() { allocate(); }
    private native void allocate();

    /** This method is invoked by C++ when a ConnectionError event fires. */
    @Name("operator()")
    public abstract void call(@ByRef Mavsdk.ConnectionError err);
}

But I keep getting compile time errors saying:

definition of type 'ConnectionErrorCallback' conflicts with type alias of the same name [arm64-v8a] 

How do I get around this by telling the builder that I don't want to have this generated this way?

I've tried just referring to the type without the alias and get another error:

  template specialization requires ‘template<>

which sucks because I try to define the template like it says to in the wiki but the InfoMapper doesn't take into consideration my hand written class.

Any help?  

[Samuel Audet]

If I understand correctly what you're looking for, we should be able to
get this working with something like
infoMap.put(new
Info("std::function<void(ConnectionError)>").pointerTypes("ConnectionErrorCallback").define());

[Rod B]

Thank you! That worked great.

I hope I'm not a bother but is there an exception for templates that take no args? My C++ skills aren't great so I could be wrong in describing this but I have another typedef

using ConnectionHandle = Handle<>;

and for some reason it won't generate correctly? Anytime the ConnectionHandle alias is mentioned in the C++ source code it is picked up by the parsers and some default stuff gets generated but even if I use .define() or .PointerTypes() the best that happens is that the garbage is replaced with what I describe in .PointerTypes but the java peer is never generated?

[Samuel Audet]

That probably just means that the file where the Handle template is
defined isn't in the include list.

[Rod B]

Crazy enough, it is being included? I'm gonna try using the logger to see if I can figure out what the issue is

[Samuel Audet]

I see, it's probably having problems picking up the default argument.
JavaCPP's parser isn't good enough for that. You'll probably need to do
something like
infoMap.put(new
Info("Handle<theDefaultArgument>",
"Handle<>").pointerTypes("ConnectionHandle").define());

[Rod B]

What if Handle doesn't have a default argument?

#pragma once

#include <cstdint>

namespace mavsdk {

template<typename... Args> class FakeHandle;
template<typename... Args> class HandleFactory;

/**
* @brief A handle returned from subscribe which allows to unsubscribe again.
*/
template<typename... Args> class Handle {
public:
Handle() = default;
~Handle() = default;

/**
* @brief Wheter handle is valid
*
* @return true if handle is valid
*/
bool valid() const { return _id != 0; }

bool operator<(const Handle& other) const { return _id < other._id; }
bool operator==(const Handle& other) const { return _id == other._id; }

private:
explicit Handle(uint64_t id) : _id(id) {}
uint64_t _id{0};

friend FakeHandle<Args...>;
template<typename...> friend class HandleFactory;
};

} // namespace mavsdk

[Samuel Audet]

Ah, a variadic template. That's not something that's supported too well
either, but it should work. If it doesn't please open an issue with a
simple example to reproduce it:
https://github.com/bytedeco/javacpp/issues