Re: receiving message from QSYSOPR msgq

Vern Hamberg wrote:

I've found the technique of looping through message keys to be
very effective. This has long been recommended by IBM. I've used
it to bracket a process where messages can come from diverse
sources. Article is entitled "Receiving Messages from the Joblog
after the Program Is off the Callstack", number 28942837. Would
that be useful in this scenario?

While that algorithm is effective for receiving program messages from programs no longer on the stack, the QSYSOPR is a non-program message queue; i.e. there is little value in mimicking that algorithm explicitly, because no program message queues are involved. That algorithm is mimicked however, however just by using the command parameter special values; no different than for when all of the programs [program message queues] remain active on the stack.

The non-program message queue can be searched simply by requesting the RCVMSG using MSGTYPE(*LAST|*FIRST) MSGKEY(&FrstRcvd), followed by a loop using the MSGTYPE(*PRV|*NEXT) MSGKEY(&PrevRcvd). The value in reference to the noted KB article is the concept of obtaining an original message key variable in order to know when to exit from the loop when reading backward, or where to start other than *FIRST when reading forward. However the /sender/ information can provide a date\time value against which compare could be used instead, to decide when to exit the loop. Or even all messages on the queue could be received every time, where the exit is only by compare to the expected MSGID value; of course that is a bad idea if the system might have gone into restricted state prior to this request before either an implicit CLRMSGQ *SYSOPR by the OS for IPL or an explicit request since the last ENDSYS.

Regards, Chuck