Well, philosophically I think that unless you have at least repeatable read
isolation or are basing a conclusion on only one read your results are
logically suspect.

However, religious wars aside, I'm mostly suggesting that if we run tests
with long explicit transactions in both interclient and jbird we can find
out how much of the performance difference is due to transaction handling.
It it is a lot, we have more incentive to try to speed up the autocommit
case. If not, we can look elsewhere.

thanks
david jencks

On 2002.10.16 12:32:19 -0400 Carsten Schäfer wrote:
> I think for inserts and updates you are right.
> But why should i use explicit commit for queries ?
> I always use autocommit mode when I'm only reading values from the db.
>
> gruß
> Carsten
>
> ----- Original Message -----
> From: "David Jencks" <davidjencks@...>
> To: <Firebird-Java@yahoogroups.com>
> Sent: Wednesday, October 16, 2002 6:17 PM
> Subject: Re: [Firebird-Java] Driver Speed Tests
>
>
> > IMO our driver should be faster and have lower latency.
> >
> > I haven't looked at why we are slower yet. My first guess is that we
> spend
> > a lot of time dealing with autocommit. This test appears to use
> > autocommit. For each operation that contacts the database, we are
> making
> > at least 3 calls: start tx, do work, commit tx.
> >
> > There may be a firebird native autocommit setting I haven't found which
> > would reduce this to one call.
> >
> > In any case (IMO) autocommit is not used in real work and we should run
> the
> > tests with larger amounts of work/transaction and explicit transaction
> > control.
> >
> > thanks
> > david jencks
> >
>
>
>
> To unsubscribe from this group, send an email to:
> Firebird-Java-unsubscribe@yahoogroups.com
>
>
>
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
>
>
>
>
>