On Feb 17, 2008, at 5:40 PM, Oliver Jowett wrote:
> (1) Map query timeout to server-side statement_timeout. Add a
> timeout parameter to QueryExecutor methods. The protocol layer
> remembers the current setting of statement_timeout and issues an
> appropriate "SET statement_timeout" as necessary before submitting
> each query.
>
> Advantages: Gives nice failure characteristics (query is cancelled,
> connection remains usable)
> Disadvantages: Doesn't help with anything but slow queries on the
> server side, relies on server-side query cancellation due to timeout
> happening reasonably promptly. Client code that sets
> statement_timeout itself can confuse it.
i think this is a very reasonable approach. it seems like it should
handle the majority of the cases.
-pete
--
(peter.royal|osi)@pobox.com - http://fotap.org/~osi