Re: RCAC - Overriding Masked Values - was: Another FieldProc question

No quite the opposite. It makes it less dangerous. Because the old
school guy is not going to do RCAC except at gunpoint. Therefore the DBA
will. And the DBA will get the constraint right.

I will say this though, the classic example of a SSN mask to 'xxx-xx-####'
is easy to think of way to handle with a constraint. Masking numerics is
a little trickier to handle with a constraint to not change a masked
value.. And this is where the old school guy may get in the way. For
example, if you're not authorized to just the salary column of an
executive and it is then masked what do you mask it as? Null? Well, the
old school guy will just replace that with zeros. Therefore if your
constraint says if null then don't change the end result, you're toast.
This may be where you need to create a test program with some of the logic
from a good representative of an update program. Or maybe handle with a
trigger instead of a constraint. As triggers are not passed the masked
value, they are passed the real values in the before/after buffers.

I do agree that if people only updated their intended columns by using
update myRcdFmt %fields(colA, colE, colG)
it would make things easier.

Rob Berendt