I am in the process of implementing a bridge support for a new middleware component in openMAMA. During implementation I have a need to return a sensible mama_status code for a generic bridge error.

From: Mike Schonberg <mschonberg@...> These changes ensure that we hold the lock while opening and closing payload and messaging brdiges. Additionally, we store the shared library handles so we

From: Mike Schonberg <mschonberg@...> Added a reference count to mamaBridgeImpl struct. mama_start() and mama_stop() increment and decrement this reference count to determine when initialization

From: Mike Schonberg <mschonberg@...> These methods take an additonal unsigned int* parameter that upon return contains the the reference count for mama_openXXX() and mama_closeXXX(). This allows

From: Mike Schonberg <mschonberg@...> Improved the reference counting logic and added locking around mama_open() and mama_close(). The change ensures that all initialization (one-time) and

From: Mike Schonberg <mschonberg@...> This is in preparation for adding improved reference counting and locking around mama_open(), and mama_close() to enable them to be called safely

From: Mike Schonberg <mschonberg@...> This method allows bridges to pass middleware specific details regarding errors and events. Signed-off-by: John Gray <jgray@...> ---

From: Mike Schonberg <mschonberg@...> Currently, the FT setup code uses the hard-coded transport name, "ft", to read properties from mama.properties despite the fact that the caller supplies

From: Mike Schonberg <mschonberg@...> This avoids a potentional sgmentation fault in the event that an uninitialized callback gets invoked. Signed-off-by: John Gray <jgray@...> ---

From: Mike Schonberg <mschonberg@...> When amessage arrives with a status of POSSIBLY_STALE, verify that the message type is MAMA_MSG_TYPE_MISC. This behavior is consistent with OpenMAMA

From: Mike Schonberg <mschonberg@...> This method refactors repeated logging statements and make the source easier to follow. Signed-off-by: Micheal Schonberg <mschonberg@...> ---

From: Mike Schonberg <mschonberg@...> Renamed listenerMsgCallbackImpl_invokeErrorCallback to listenerMsgCallbackImpl_invokeErrorCallback because it is an internal method. The method is also

From: Mike Schonberg <mschonberg@...> Middleware bridges pass this status to the mamaSubscription onError callback when the middleware detects a gap:

From: Mike Schonberg <mschonberg@...> The existing logic contained a race condition which would not consistently log the attempt to dereference a queue with a zero reference count. The

From: Mike Schonberg <mschonberg@...> Free memory and fail if realloc fails in propertiesImpl_addKey(). When we replace a value in the hash table in propertiesImpl_addProperty(), free the

From: Mike Schonberg <mschonberg@...> The C++ wrappers require mama/types.h to be included from mamainternal.h in order to build properly. Signed-off-by: John Gray <jgray@...> ---

From: Mike Schonberg <mschonberg@...> Avoids potential namespace conflicts as the hash functions are now used outside of common. Signed-off-by: John Gray <jgray@...> ---

From: Mike Schonberg <mschonberg@...> Added method to disable refreshes for a transport. This method must be called after mamaTransport_allocate() but before mamaTransport_create(). If

This list has been pretty quiet over the past couple of months. That does not mean that we have not been busy. In fact, the list has been quiet because we have been fully engaged preparing the

Fault tolerance for multicast. ? ft.patch Index: ft.c =================================================================== RCS file: /cvsroot/products/mama/c_cpp/src/c/Attic/ft.c,v retrieving