On Sun, 7 Apr 2013 15:58:35 -0400, Phil <[email protected]> wrote:
>Clearly this is another case of RTFM. I missed JSTCB=YES on my ATTACH. > >The manual does say 'one and only' jobstep task, but > > . must be ... a job step task > . should be ... one and only JOBSTEP TCB > >must. should. "a job step task". JOBSTEP TCB. > >Subtleties. "a" implies there could be more than one? Yes, there can be more than one. Well, there is always more than 1, but let's pretend it said "jobstep TCB running a user program"; all the others are at the initiator and above. That first user-specified jobstep TCB can, if it's running authorized, attach additional jobstep TCBs, as long as all of its direct children are jobstep TCBs. But you _should_ not do that with the TMP, as it wants to be the only user-specified jobstep TCB. That is, in my interpretation, it wants to be running in the TCB attached by the initiator. So if you have another program invoking the TMP it _should_ LINK or XCTL to it, so it's in the "proper" TCB. I have no idea whether it really makes a difference or not. -- Walt
