On Feb 25, Zardosht Kasheff wrote:
> Can you please elaborate on the problems?
> Normally, with Insert ignore, a disk seek is required to verify
> uniquness of the element (and thus the return of an error). If I do
> not need to return an error, the disk seek is unnecessary.
I think I'm missing something here.
If you don't know if the element is unique or not what do you do with
the row data ? Is the new row inserted into the table ? Do you check
uniqueness in the background ? Or you just insert it and ignore later
during table reads ?