Tom Lane wrote:
> Bruce Momjian <pgman@candle.pha.pa.us> writes:
> > Tom Lane wrote:
> >> Because we want commits/rollbacks to be counted if any of them are on.
>
> > Why do we want commits/rollbacks counted if we only have command string
> > enabled?
>
> Why not? Those counts are not either "tuple level" or "block level"
> operations; the fact that the implementation sends them in the same
> messages doesn't mean that there is any association in the user's eye.
> Barring making a fourth GUC variable to control them (which seems like
> overkill), I think it's a reasonably sane definition to say "we count
> these if any stats are being collected". Doing what you propose would
> simply expose an irrelevant implementation detail to users.
OK. Don't we need to document this somewhere?
> > The !(x || y) construct is really ugly and I will fix that in a simple
> > commit now.
>
> I can't agree with you on that opinion, either.
Oops, done.
The good news is I found out why stat_command_string is causing such a
large performance hit. I will post tonight or tomorrow on it.
--
Bruce Momjian | http://candle.pha.pa.us
pgman@candle.pha.pa.us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073