axis-java-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Deepal Jayasinghe" <>
Subject Re: [Axis2] Possible bug in the AxisEngine.resume() method
Date Mon, 19 Dec 2005 07:47:25 GMT
Hi Chamikara;

I am not sure I am giving the right answer, any way the reason behind treating Message reciver
seperately is the following

 say that you add message receiver as the end of the operation chain and you just call msgCnx.invoke();
, what if some one change the handler chain at the runtime to add a handler after the message
receiver. To avoid that kind of scenario we treat Message Receiver separately.

~Future is Open~

  ----- Original Message ----- 
  From: Chamikara Jayalath 
  Sent: Thursday, December 15, 2005 8:29 PM
  Subject: Re: [Axis2] Possible bug in the AxisEngine.resume() method

  Hi Glen,

  Since TransportSender already extend the Handler interface it can be easily attached to
the end of the handler chain. But MessageReceiver is a seperate interface and this does not
extend Handler.

  Are you suggesting that the MessageReceiver should extend the Handler interface. (also this
will remove the need for a  MessageReceiver.receive() method  since the invoke() method from
the Handler interface can be directly used).

  (I don't know weather there is a special reason for MessageReceiver not being a Handler
in the current implementation)

  Thank you,

  On 12/16/05, Glen Daniels <> wrote:

    >     So the end result is a single execution chain which pauses/resumes
    >     simply by indexes, updates itself at operation dispatch, and no extra
    >     logic to split receiving/sending.
    >     Sound reasonable? 
    > +1

    OK, cool.  I'm more than willing to do this work next week (am on
    vacation as of tonight for the weekend), but assuming others are good
    with this, please feel free to commit the patch and/or the refactoring 
    before then if you wish.


View raw message