Re: Custom MONITOR handling

[Barbara Morris <bmorris@xxxxxxxxxx>]

Steve Richter wrote:
> 
> I dont follow the problem of fitting rpg exception handling in with the
> whole OS/400 messaging protocol.
> 
> Here is a way that the "Cee" APIs can be used to monitor for specific
> message ids and either halt or promote those that are not monitored for.
> ...
> Is this not an acceptable way to handle exceptions in proc level rpg code?
> ...

Steve, using CEEHDLR can be dangerous if you handle the exception so
that control returns to the next statement.  The reason is that the next
statement is not the next high-level-language statement - it's the next
MI statement.  This could take you back into the middle of the
processing of your RPG statement and possibly cause data corruption. 
For example, say you are handling this statement:
    x = y + z
where z is based, and you get a pointer-not-set exception for z.  Your
CEEHDLR handler handles the MCH3601 and control returns to the add
statement and then on to the store to x.  What does it add?  Who knows.

This is a made-up scenario - I have no idea how that particular scenario
would work.  But even if you established that it worked ok, it might not
continue to work ok.  The system doesn't guarantee upward compatible
behaviour for the results of ignoring an exception.  But besides that,
the compiler might generate the code slightly differently even across
PTFs which could cause the ignore-exception behaviour to change.

(Exactly the same kind of problem can happen with RPG III programs that
IGNDECERR.  RPG III generates an ignore handler which causes the program
to keep going with sometimes partially updated fields (!).)