Re: More fun with native RLA SETLL on an SQL view with a WHERE clause

["James H. H. Lampert" <jamesl@xxxxxxxxxxxxxxxxx>]

CRPence wrote:

>    The pointer stored in the DMEPT has to come from the *SEPT, so 
> just update the QINSEPT before the file is opened.  I would make the 
> intercept program a CLP that does nothing more than a TFRCTL to the 
> OS DB program, in order to avoid impacting any other work that must 
> be done while debugging.  Or after the open, but before the I\O 
> request, the DM EPT could be updated in the ODP.

Hack the QINSEPT?!? On a production box?!? Uh, even if I knew how (I had 
to read the aforementioned tutorial from Leif, just to know how to dump 
it!), it's not something I'd care to risk.

Hacking the ODP would be a little more reasonable, IF I had the 
slightest idea how to do that in an ILE program.

The first recompilation only yielded a few ideas that DIDN'T work:
If I changed the option list on a .DMCGETD (RPG SETLL) to "Read First," 
the program crashed.
If I set the target RRN to zero or a negative value, the OPM RPG test 
got nothing, and the MI app got the known-excluded first record.


But I've got another recompilation cooking right now (it just finished, 
but I've been off-duty for 5 minutes already!), for tomorrow morning, 
that's got the option list for .DMCGET (RPG READ) opened up for 
manipulation in the debugger.

--
JHHL