Yes, and while I might not be at liberty to publish *EVERYTHING* I do, I
will definitely say that the option that appears the most practical,
given that there's no easy way of knowing the call-stack-depth from
whence the message is being sent, relative to the caller of the service
program, is to simply duplicate the QMHSNDPM call already in
JDBCR4.jni_checkError(), sending a copy of the message (which is of type
DIAG) to the external message queue (where anybody can receive it).