Zeugswetter Andreas SB <ZeugswetterA@wien.spardat.at> writes:
> this. Given that cursors (are supposed to) support FETCH BACKWARDS,
> I really don't see why they shouldn't be expected to handle ReScan...
>> I thought only scrollable cursors can do that. What if cursor isn't
>> scrollable? Should it error during the execution?
> In PostgreSQL, all cursors are scrollable. The allowed grammar keyword is
> simply ignored. I am actually not sure that this is optimal, since there
> are a few very effective optimizations, that you can do if you know, that
> ReScan is not needed (like e.g. not storing the result temporarily).
It's worse than that: we don't distinguish plans for cursors from plans
for any other query, hence *all* query plans are supposed to be able to
run backwards. (In practice, a lot of them don't work :-(.) Someday
that needs to be improved. It would be good if the system understood
whether a particular plan node would ever be asked to rescan itself or
run backwards, and could optimize things on that basis.
regards, tom lane