At 01:36 AM 07-11-02 -0800, you wrote:

>Thanks, Helen. Another question:
>
>Here's what I'm actually doing in my loop:
>
>Grab the key from the source data record.
>Do a select to see if the key already exists.
>If (key doesn't exist)
> Insert it
>else
> Update it
>
>Now, with the scenario that you outlined below (Using the transaction),
>what will happen if the source data has the same key twice? If I'm not
>committing after each insert, will the next select accurately know whether
>or not the key already exists (if it was previously inserted but not
>committed)?

The next insert will see anything that has already been inserted within the
same transaction. Your transaction can see everything that it has done so
far...

If you have to test for existence in a batch insert, you should *not* be
doing this in the client. I'm curious to know how you are doing
conditional insert/update in batch in a single DSQL statement.

>One other thing: If I have a hard-coded value that I'm inserting into EVERY
>record, is there any performance benefit to hard coding it into the insert
>statement, as opposed to leaving it as a parameter that I set the same
>every time?

Of course. Don't make it a parameter at all, but a constant in the SQL.

Helen