OSA reference

BMIP request processor

Each OSA can control how the implementation of BMIP requests is actually performed. This is performed by informing the Server API about an OSAs request processor.

NOTE: The implementation of a request processor is optional, and should only be done if the OSA needs some special functionality such as synchronization, advanced scoping or first pass validation.

By providing a request processor, the OSA is taking over control of processing the BMIP requests from the Server API. In the very least this entails:

running a filter via OFEvaluateFilter(3tlib) on the object instance(s).
executing the specified operation on the object instance(s) via OFEvaluateOperation(3tlib).
returning a BMIP response to the client via OFReturnBmipResponse(3tlib).

See ``Example request processors'' for both Tcl and C/C++ implementations. The request processor is the only place an OSA class can implement advanced scoping or synchronization. In addition to the minimal operation set, these features can be added:

By passing the request processor address/handle to the OFBinding(3tlib) procedure, the Server API is informed that the routine should do the scheduling of calling OSA functions rather than the Server API. Otherwise a NULL (C/C++) or lack of argument (Tcl) should be passed, indicating that the Server API should perform the scheduling.