Thanks a lot for your help. I've implemented the successive "FETCH n" commands approach and it works a treat.
-James
Tom Lane wrote:
> Oliver Jowett <oliver@opencloud.com> writes:
>
>>It might even work to execute FETCH FORWARD ALL for the cursor with a
>>non-zero fetchsize (and let the driver manage incremental fetches) but
>>I'm not sure what the behaviour of a non-zero row limit on a FETCH query
>>is at the protocol level..
>
>
> [ digs in code... ] It looks like the behavior would be to (a)
> materialize the entire result of FETCH FORWARD ALL inside the backend,
> then (b) hand back the number of rows you requested. Probably not what
> you want :-(. I'd suggest issuing successive "FETCH n" commands and
> ignoring the protocol-level limit feature.
>
> regards, tom lane
>
> ---------------------------(end of broadcast)---------------------------
> TIP 8: explain analyze is your friend
>
>
>