A client is running build .30305, which I believe is the latest ..
A program (run daily, untouched since 9/2009) does an insert into
a table using a combination of columns and variables. The syntax is:
INSERT INTO sales +
(invoiceno, trandate, trantime, accountno, billaccount, brate) +
SELECT invoiceno, .#DATE, .#TIME, accountno, .vmacctno, .vBillRate +
FROM tempsales WHERE .....
The 2 variables referenced (vmacctno and vbillrate) are predefined, and are
populated by a "select into". However, the vBillRate could be null. In
that
case, I see it in my variable list as
vBillRate REAL with no value
I swear this must have happened before (but maybe not), but when vBillRate
is null,
the insert fails and I get the message
"An expression cannot end with binary operator"
If I replace the vBillRate with a 0 , then the insert works. Whileopt is
OFF, Zero is ON.
My null setting is the default -0-. If I change the null setting to ' '
(blank space)
right before the insert, the insert still fails but with:
"incorrect number of values for this table"
So now I'm pretty worried about other places I might have a null variable
being
inserted... What do other programmers do? Is this bad programming?
Should
each variable be populated with something? What if you want a number value
to be truly null, and not a 0?
Karen
