Re: detecting "bad data" using an sql statement

On 1/17/2018 11:05 AM, Jay Vaughn wrote:

However, the data currently exists in our PF's and our applications which
are 99% rpg pgms do not blow up.

Why is this?

Here, I see the following reasons:

1) Program described I / O-specs.
2) Programs compiled with IGNDECERR / FIXDECERR.
3) Not using the 'bad' column.

With RLA, if a column is RNF7031 - Unreferenced, it won't throw a
decimal data error if you CHAIN to the record. The field is, well, not
referenced by the program, and therefore the compiler doesn't try to map
the database file buffer into the (unreferenced) RPG fields. No buffer
transfer, no validation error.

With SELECT *, that same column in the same file with throw a decimal
data error. I think the database manager validates it before RPG gets
to it. This is probably over simplified, but it's a mental model that
works well enough for me to understand when I get decimal data errors
and when I do not.