Pluggable_Messaging.h

Go to the documentation of this file.

// -*- C++ -*-

//=============================================================================
/**
 *  @file     Pluggable_Messaging.h
 *
 *  Pluggable_Messaging.h,v 1.32 2006/04/20 12:37:17 jwillemsen Exp
 *
 *   Interface for the TAO pluggable messaging framework.
 *
 *
 *  @author  Balachandran Natarajan <bala@cs.wustl.edu>
 */
//=============================================================================

#ifndef TAO_PLUGGABLE_MESSAGING_H
#define TAO_PLUGGABLE_MESSAGING_H

#include /**/ "ace/pre.h"

#include "tao/TAO_Export.h"

#if !defined (ACE_LACKS_PRAGMA_ONCE)
# pragma once
#endif /* ACE_LACKS_PRAGMA_ONCE */

#include "tao/Basic_Types.h"

ACE_BEGIN_VERSIONED_NAMESPACE_DECL
class ACE_Time_Value;
ACE_END_VERSIONED_NAMESPACE_DECL

TAO_BEGIN_VERSIONED_NAMESPACE_DECL

namespace CORBA
{
  class Exception;
}

class TAO_Target_Specification;
class TAO_Pluggable_Reply_Params;
class TAO_Pluggable_Reply_Params_Base;
class TAO_Transport;
class TAO_Operation_Details;
class TAO_Target_Specification;
class TAO_OutputCDR;
class TAO_Queued_Data;
class TAO_GIOP_Fragmentation_Strategy;

// @@ The more I think I about this class, I feel that this class need
// not be a ABC as it is now. Instead we have these options
// (1) Parametrize this class with the "Messaging Object". Then the
//     implementations  can just redirect the request to the class
//     that it has been specialised with
// (2) Use a bridge pattern here ie. the interface exposed by this
//     class will redirect the request to the implementation which
//     would be "right messaging" object.
// Both of them feasible. If we are not going like templates much, (2)
// is seems to be a better option.
// I am going to take a closer look on this after my MS defense -
// Bala

/**
 * @class TAO_Pluggable_Messaging
 *
 * @brief Generic definitions Messaging class.
 *
 * This interface tries to define generic methods that could be
 * different messaging protocols
 */
class TAO_Export TAO_Pluggable_Messaging
{
public:
  /// Dtor
  virtual ~TAO_Pluggable_Messaging (void);

  /// Write the RequestHeader in to the @a cdr stream. The underlying
  /// implementation of the mesaging should do the right thing.
  virtual int generate_request_header (
      TAO_Operation_Details &op,
      TAO_Target_Specification &spec,
      TAO_OutputCDR &cdr) = 0;

  /// Write the RequestHeader in to the @a cdr stream.
  virtual int generate_locate_request_header (
      TAO_Operation_Details &op,
      TAO_Target_Specification &spec,
      TAO_OutputCDR &cdr) = 0;

  /**
   * Write the reply header
   * The reply header is a part of any messaging protocol. The
   * messaging protocol implementation would implement the way the
   * reply header is written.
   */
  virtual int generate_reply_header (
      TAO_OutputCDR &cdr,
      TAO_Pluggable_Reply_Params_Base &params) = 0;

  /**
   * Write the locate reply header
   * The locate reply header is a part of any messaging protocol. The
   * messaging protocol implementation would implement the way the
   * locate reply header is written.
   */
  virtual int generate_locate_reply_header (
      TAO_OutputCDR &cdr,
      TAO_Pluggable_Reply_Params_Base &params) = 0;

  virtual int generate_fragment_header (TAO_OutputCDR & cdr,
                                        CORBA::ULong request_id) = 0;

  /// Format the message in the @a cdr. May not be needed in
  /// general.
  virtual int format_message (TAO_OutputCDR &cdr) = 0;

  /// Do any initialisations that may be needed.
  virtual void init (CORBA::Octet major,
                     CORBA::Octet minor) = 0;

  /// Parse the details of the next message from the @a incoming
  /// and initializes attributes of @a qd. Returns 0 if the message
  /// header could not be parsed completely, returns a 1 if the message
  /// header could be parsed completely and returns -1 on error. As the
  /// parsed data is stored directly in @a qd, no state must be stored
  /// in instance of implementation.
  virtual int parse_next_message (ACE_Message_Block &incoming,
                                  TAO_Queued_Data &qd,        /* out */
                                  size_t &mesg_length) = 0;   /* out */

  /// Extract the details of the next message from the @a incoming
  /// through @a qd. Returns 0 if the message header could not be
  /// parsed completely, returns a 1 if the message header could be
  /// parsed completely and returns -1 on error.
  virtual int extract_next_message (ACE_Message_Block &incoming,
                                    TAO_Queued_Data *&qd) = 0;

  /// Check whether the node @a qd needs consolidation from @a incoming,
  /// @r 0 on success, -1 on error
  virtual int consolidate_node (TAO_Queued_Data *qd,
                                ACE_Message_Block &incoming) = 0;

  /// Parse the request message, make an upcall and send the reply back
  /// to the "request initiator"
  virtual int process_request_message (TAO_Transport *transport,
                                       TAO_Queued_Data *qd) = 0;


  /// Parse the reply message that we received and return the reply
  /// information through @a reply_info
  virtual int process_reply_message (
      TAO_Pluggable_Reply_Params &reply_info,
      TAO_Queued_Data *qd) = 0;


  /// Generate a reply message with the exception @a ex.
  virtual int generate_exception_reply (
      TAO_OutputCDR &cdr,
      TAO_Pluggable_Reply_Params_Base &params,
      CORBA::Exception &ex) = 0;

  /// Is the messaging object ready for processing BiDirectional
  /// request/response?
  virtual int is_ready_for_bidirectional (TAO_OutputCDR &msg) = 0;

  /// Reset the messaging the object
  virtual void reset (void) = 0;

  /// Header length
  virtual size_t header_length (void) const = 0;

  /// Fragment header length
  virtual size_t fragment_header_length (CORBA::Octet major,
                                         CORBA::Octet minor) const = 0;

  /// Accessor for the output CDR stream
  virtual TAO_OutputCDR &out_stream (void) = 0;

  /// Consolidate newly received fragment with previously arrived
  /// associated fragments to achieve consolidated message.  All
  /// fragments having been received previously are being managed
  /// within implementation.  If reliable transport is used (like TCP)
  /// this operation will be invoked with fragments being received
  /// partially ordered, last fragment being received last. Otherwise
  /// If un-reliable transport is used (like UDP) fragments may be
  /// received dis-ordered, and must be ordered before consolidation
  /// within implementation.  @return 0 on success and @a con_msg
  /// points to consolidated message, 1 if there are still fragmens
  /// outstanding, in case of error -1 is being returned. The
  /// implementation is responsible to release @a qd.
  virtual int consolidate_fragmented_message (TAO_Queued_Data *qd,
                                              TAO_Queued_Data *&con_msg) = 0;

  /// Discard all fragments associated to request-id encoded in
  /// cancel_request.  Transport implementaion guarantees that this
  /// operation will never be invoked // concurrently by multiplpe
  /// threads nor concurrently to consolidate_fragmented_message
  /// @return -1 on failure, 0 on success, 1 no fragment on stack
  /// relating to CancelRequest.
  virtual int discard_fragmented_message (const TAO_Queued_Data *cancel_request) = 0;

  /// Outgoing GIOP message fragmentation strategy.
  virtual TAO_GIOP_Fragmentation_Strategy * fragmentation_strategy (void) = 0;
};

TAO_END_VERSIONED_NAMESPACE_DECL

#include /**/ "ace/post.h"

#endif /*TAO_PLUGGABLE_MESSAGING_H*/

---