RE: Dynamic select, null dates, -183, COALESCE, CRTSQLRPGI, DATFMT(*ISO)

[Buck Calabro <Buck.Calabro@xxxxxxxxxxxx>]

>Well, so much for progress; I have tons of null dates.
>SQL reminds me of free-form I-specs, replete with
>opportunities for swapped data elements, etc.  Seems
>to me there's an opportunity for the RPG developers to
>clean this interface up a little...

Hi Reeve!
I think they WANT to, but there's some internal debate over whether the DB2
folks or the compiler folks should pay for it.  Budgets rule us all.

Indicator variables aren't too bad; they let you preserve the 'nullness'
flag as opposed to masking it with unlikely data (which is what coalesce
does).  The problem I've run into with coalesce is that no matter what
unlikely data I choose, somebody eventually decides to use it for their own
nefarious purposes.  Not to mention that it's hard to tell exactly how many
of these 'special' values are floating around in the code base.  I have a
true story about a payroll system where the owner's employee number
triggered some special deductions.  He retired and years later his employee
number was re-issued.  Somebody got a very pleasant first pay check!  I have
been wary of special values in code ever since.

Another approach to look at for dynamic SQL is the CLI.  The iSeries-toolkit
has some good examples which is a Good Thing, since the manuals are written
for ODBC vendors.  Well, it seems to me they are.
http://www.iseries-toolkit.org

  --buck, still working on weaning myself from READE...