Session A501 - Should Applications be Allowed to Multi-task?

March 1-6, 1992

This session very possibly had the worst session title in recent SHARE history. Two IBMers from San Jose came to discuss technical problems that IBM has to address when doing error recovery for multitasked applications. Instead of the usual passive SHARE audience, they attracted an unusually rabid crowd of foaming systems and applications people, who believed (based on the session title) that IBM was planning to remove multitasking capabilities from new versions of software. We freaked out.

Questions we asked of the IBMers ricocheted back and forth between them in a kind-of-embarrassed, kind-of-irritated, what-have-we-gotten-ourselves-into? tap dance. It took most of 40 minutes before we discovered that IBM doesn't want to restrict multitasking at all!

Then it took five minutes for this inarticulate crew to tell us that their real problem is error recovery in multitasking systems.

I don't think they thought the problems through much before coming to SHARE to expose themselves. They weren't able to tell me why they weren't concerned with APPC-style multitasking (tasks running on different processors) but cared about tightly-coupled systems (tasks running on a single machine).

I think their real problem is: how do we implement multitasking in an SAA world? Tasks, threads and processes are all different things, and they also mean completely different things on different platforms.

Back to session index
Back to index of SHARE meetings
Read the disclaimer