Re: Uncaught Errors in Interactive Programs and the Users that Ignore the Errors

[CRPence <CRPbottle@xxxxxxxxx>]

  But not having coded a global handler just means that the 
language's global handler is invoked... which it seems, may be 
exactly the "problem" the OP wants to address.  Thus to prevent the 
CPA0701 inquiry [or similarly default exception handling inquiry 
processing, specific to the language], a programmer-defined global 
monitor would have to supersede the implicit language-specific 
global monitor.

  So although MONMSG CPF0000 EXEC(*N) is not a good global monitor 
because that is just a "global ignore" of errors, actually coding a 
handler by specifying EXEC(do_something) can offer great benefit. 
The "do_something" could even effect an inquiry, but not so easily 
effect a retry.  As a programmer I prefer the 'R' retry, as alluded 
to enable correction then retry, but a typical "user" will have 
limited capability to make such decisions or even correctly perform 
corrective before retry when they do.

  However, if the user jobs are forced to run with inquiry message 
reply handling of something other than *RQD, then they may not even 
be offered an opportunity to answer an inquiry.  For example all of 
the HLL generic exception inquiry messages could be defaulted to 
reply with the 'D' dump option, either by message default or the 
default reply value.  That may be something the OP should consider, 
if one of the valid reply values might offer a better outcome than 
if the decision is left to the user; e.g. Enter pressed, thus 
sending *N or the default as the reply value.

  Another option, but one I would not actually recommend, is the 
inquiry message exit program.  That exit could recognize such a 
generic HLL handler inquiry, and perform the notifications & 
gathering of details about the failure; optionally choosing how to 
respond for the user.

Regards, Chuck

rob@xxxxxxxxx wrote:
> Users can be vexing, I understand. But I often loathe generic
> error handlers, like wrapping all your RPG logic with one bit
> monitor. Or putting a MONMSG CPF0000 as the first line in your
> CL programs.  Especially in CL. Because if I see the message I
> can often patch something and issue a retry. Use a global and
> it confuses the retry.