Zeugswetter Andreas SB <[EMAIL PROTECTED]> writes:
>> Right. So what do you think about a hint that takes the form of a SET
>> variable for the fetch percentage to assume for a DECLARE CURSOR?
> Since we don't have other hints that are embedded directly into the SQL
> that sounds perfect.
> The not so offhand question for me is whether to use this percentage
> for non cursor selects also. Imho both should (at least in default)
> behave the same.
Not at all, since in a non-cursor select you *must* retrieve all the
data. I can't see any reason to optimize that on any other basis than
total execution time.
regards, tom lane
