Re: RCVMSG hell...

[Buck <kc2hiz@xxxxxxxxx>]

On 8/19/2011 1:27 AM, Vern Hamberg wrote:

> Otherwise, there's a method of getting all messages by key between the
> start of a process and the end - IBM put out  KB on it, maybe look for
> how to find all messages from a job log. That might catch these messages
> that aren't available through standard typical ways.

The method is easy enough.  You send a place holder message at the
beginning of your process and save the message key.  You send another
message at the end and save that key.  Then do a RCVMSG for every
message key in that range.

sndpgmmsg 'start save' topgmq(*same) keyvar(&key_start)
rmvmsg    msgkey(&key_start)

/* do a big pile of stuff */


/* big pile finished... did any errors occur? */
  sndpgmmsg 'stop save' topgmq(*same) keyvar(&key_stop)
  rmvmsg    msgkey(&key_stop)

  /* calculate the first possible diagnostic message */
  chgvar %bin(&msgkey 1 4) (%bin(&key_start 1 4) + 1)

get_msg:
  /* get the diagnostic message */
  RCVMSG PGMQ(*SAME (*)) +
         MSGTYPE(*DIAG) RMV(*no) MSGDTA(&MSGDTA) +
         MSGID(&MSGID) msgkey(&msgkey)
  monmsg cpf2410 exec(do) /* msg not found */
    rcvmsg msgtype(*excp) rmv(*yes) /* delete unwanted msg */
    goto next_msg
  enddo

  /* if empty, all done */
  if (&msgid *eq '       ') goto exitpgm

/* process the message */

next_msg:
  /* increment the message key */
  chgvar %bin(&msgkey 1 4) (%bin(&msgkey 1 4) + 1)
  if (&msgkey *lt &key_stop) goto get_msg

exitpgm: