Re: PlainSocketImpl.java PlainSocketImpl.c and my own socket implementation

Still in this discussion, in the constructor of my MySocket class, I
call super(host, port) and set the SocketImplFactory passing my
factory as argument. When I create a instance the method bind is
called due to Socket implementation, but instand of the bind method
receive the host and port parameters passed when I create MySocket
instance, the method bind receives host=0.0.0.0 and port=0, but right
before send these values as arguments (host and port) I call a
System.out.println(host + port) and it is printed the values passed to
MySocket constructor. What is wrong? Please, someone to help me.
Thanks,

Leandro

2007/2/4, Leandro Melo de Sales <[email protected]>:
> Hi Steve and others,
>
>    I'm trying to create socket implementation for a transport protocol
> different of TCP and UDP (in linux), but similar to TCP, since it is a
> oriented connection protocol. I want to integrate it into the JVM. So,
> I wrote a C to Java wrapper to call native methods by using JNI and
> when I run the connect test, after create the socket and the connect
> to a remote, I can send data through out this socket. But now I want
> to extend the java socket functions to support my new socket. I'd like
> to know what is the correct process to do this. There are so many
> classes, abstract classes and interfaces which I can use, and as I
> understood, I did the job as following, but I don't know if I'm doing
> the right stuffs since after some attempts I can't create the socket
> like I wish.
>
> The steps that I'm following is:
>
> 1 - I create a class named NativeSocket.c which implements basic c
> methods, such as create socket, connect and so on. I also create a
> class NativeSocket.java which, by using JNI, I can call all native
> methods;
> 2 - I create a class named MySocketImpl which extends the SocketImpl
> abstract class and implement the abstract methods using my native
> implementation described in the step 1;
> 3 - I create I class named MySocketImplFactory which implements the
> SocketImplFactory interface, the single method createSocketImpl;
> 4 - The method createSocketImpl create and return a instance of
> MySocketImpl described in the step 2;
> 5 - I create a class named MySocket.java which extends java.net.Socket
> as follows:
>
> public class MySocket extends Socket {
>
>         static {
>                 MySocketImplFactory aMySocketImplFactory = new MySocketImplFactory();
>                 try {
>                         Socket.setSocketImplFactory(aMySocketImplFactory);
>                 } catch (IOException e) {
>                         e.printStackTrace();
>                 }
>         }
>
>         public MySocket(String host, int port) throws IOException {
>                 super(InetAddress.getByName(host), port);
>         }
>
> }
>
> 6 - I create a instance of MySocket.java like this:
>
> Socket mySocket = new MySocket("localhost", 10000);
>
> Is this the right steps to create my own socket implementation by set
> the socketimplfactory of the java.net.Socket class? When I create
> MySocket instance the method bind is called but I was expecting that
> the right method to be called is connect instand of bind. Why the bind
> method is called? BTW, what shoud I do in this bind method? I also try
> to understand the java internals implementation (the native codes) and
> in my case, do I have to worry about the SecurityManager or since I'm
> extend Socket implementation is this alredy done?
>
> Please, any clue/suggestion will be greatfully accepted.
>
> Thank you in advance,
>
> Leandro
>


----------------------------------------------------------------------
To UNSUBSCRIBE, email to [email protected]
with a subject of "unsubscribe". Trouble? Contact [email protected]