On Wednesday, 03/03/2010 at 03:53 EST, Kris Buelens <kris.buel...@gmail.com> wrote: > The cause is what in my eyes is a very old bug, it exists since the VM/XA days; > VM/SP had it right: > in the VM/XA family, when a disconnected user, without a secondary user, > causes a VM READ, it is translated into a CP READ > With SECONDARY user one can correct the problem: SEND a BEGIN command > With OBSERVER one is stuck: one can send a BEGIN, but the following VM READ is > again translated in CP READ. If that could be fixed... Or am I dreaming too > much
Indeed. CP has always assumed that a disconnected user without a SECUSER cannot possibly get a response to a VM READ. So, he changes the VM READ to CP READ and stops the virtual machine. (I speculate that it was simply a convenient shortcut to trigger the 15-minute timebomb.) With class C SEND, the above assumption is no longer true and we begin to detect a disturbance in the Force. Alan Altmark z/VM Development IBM Endicott
