[MI400] Picking up on MI debugging...

["Richard Hart" <rhart@xxxxxxxxx>]

This is a multi-part message in MIME format.
--
[ Picked text/plain from multipart/alternative ]
TO:  All and Sundry.

I read several thread segments in this mail group about debugging MI
code about a year ago.  There was nothing since!  I guess everyone but
me is now so expert on the system that debugging is unnecessary :-).

I think it is possible to create your own source level debugging program
if you really wanted to (ref:  the SRCDBGPGM parameter of the STRDBG
command).  However, the documentation states that it is only applicable
to the ILE environment.  I would imagine it is possible though; ASNA had
an OPM source level debugger for RPG before IBM came out with the
STRISDB tool.

However, one easier way would be to write an ADDBKP "front end" that
would add breakpoints and PGMVAR values for Operand 1, Operand 2 (etc.)
for each MI statement number (perhaps within a statement range) that the
user wanted  to step through; the program would then automatically stop
and display operands as it hit each MI statement.  Might be useful as a
learning tool.  The standard STRDBG interface works fine, except tedious
to type all the COBOL style variable names! (as, "PGM-PGM-NAME").

I found Leif's original GETOWNLIB program and mananged to enter and run
it successfully...sort of.  Natually, I got the "System Domain" error
when the MATPTR instruction tried to run, if the pointer in question was
for a system domain program.  Frustrating!  I guess next week I will set
up an Exception Monitor.  This (GETOWNLIB) technique is actually very
useful for solving the (somewhat) common problem of finding out which
program called the current program.  One of my previous employment sites
tried to solve that problem by some goofy CL technique that I cannot
even remember (who could?); the MI method is elegant, straightforward,
and simple (compared to the above solution).

Well, is anyone still posting to this mail group?

Rich Hart