On 30-Oct-06, at 5:25 PM, Tom Lane wrote:
> Kris Jurka <books(at)ejurka(dot)com> writes:
>> For each connection the statements BEGIN, ROLLBACK, and COMMIT are
>> prepared once and after that they are executed again and again.
>> This can
>> be very confusing when looking at the logs which only show the
>> step, but not execution. So even though you can't see them,
>> they're being
>> executed. The 8.2 release will be the first release that can
>> log this usage pattern.
> BTW, have you verified that it's actually worth the trouble to prepare
> these commands? Given that there's no planning cost for a utility
> statement, and hardly any parsing cost for one-word commands, I'm a
> dubious that the prepared-statement machinery actually results in a
> compared to just issuing simple Query messages.
Yeah, the driver needs to be taught to not prepare *everything* in v3
> regards, tom lane
> ---------------------------(end of
> TIP 4: Have you searched our list archives?
In response to
pgsql-jdbc by date
|Next:||From: Luca Ferrari||Date: 2006-10-31 10:18:55|
|Subject: Date time format out of range????|
|Previous:||From: Tom Lane||Date: 2006-10-30 22:25:26|
|Subject: Re: confused about transactions and connection pools |