Re: What's the proper way to "reraise" in a CLP?

On 6/16/2017 7:37 PM, John Yeung wrote:

I probably didn't express it very well, but the part I was most
interested in finding out was how to make the original *ESCAPE show up
again, if it turns out I can't handle this subclass. Crucially, this
*ESCAPE should put the job into MSGW status, rather than ending the
job abnormally, if the MSGW status is what would have happened had I
not embarked upon my MONMSG in the first place.

The precise definition of 'original *ESCAPE message' may make or break
the solution. In general, once your program has done a RCVMSG (or API
equivalent), the original message is consumed. The contents of that
original message are not just the message ID and text (CPF4131 'Level
check') but also which program sent the message, which program received
it, etc. I don't know of a simple way to RCVMSG and then 'mark it' as
'Not Received', so that the same, original message can cause a white
message.

(By "put the job into MSGW status" I mean another message shows up
like CPA0701.)

In general, I SNDPGMMSG a new *ESCAPE message, but with the same message
ID, text, etc.

So, nobody mentioned it explicitly, probably because it's too obvious
to everyone who knows what they are doing, but after further
experimentation it seems that the key to this is to specify
TOPGMQ(*SAME) when issuing the SNDPGMMSG that repeats the originally
caught *ESCAPE message.

Generally, I would send to *PRV; that is, to the caller. Sending it to
the program that just handled the message would result in a loop unless
there was code to detect that we've seen this one before.