> The discussion on Scheduleing an SRB into a client address space from a Cross > Memory Service Provider was very interesting. And It pronpted me to think of > the following scenario. > > Couldnt the Cross Memory Server Addresss create a Non Space Switching PC > routine that is load into MVS common. Then the client would invoke the Non > Space Switching PC Routine (residing in MVS common) to schedule the SRB. Is > this also a viable solution.
It is certainly viable. Whether or not it is a solution depends on what the problem is. My best guess is that both are perfectly legal, but overkill. If you have a long lived server address space that can anchor and manage resources appropriately then you can do pretty much anything you want. In the OP's case, they are providing a space-switch PC that is called by a unit of work running in a client address space and (for whatever reason) they are using that PC-ss to schedule an SRB back to the caller's home address space. Could they provide the same functionality with a non-switching PC? Sure. The only thing that changes is the non-switching version needs to have its code loaded in (hopefully extended) common storage. My question is more fundamental. Why use the SRB at all? The PC service caller is already dispatched on a perfectly good unit of work, has addressability to both address spaces and, while running inside the PC, is in a sufficiently God-like state to do anything you might need. CC
