Installed Firebird 1.0.3 now; seems a little bit better but still goes
up to 1.5-2 secs sometimes.
The problem with the first SQL statement intrigues me too.
Normally the first SQL statement I do on that page (after setup of the
connection and transaction) is the following:
SELECT LSTNAMEADDRESSCAT.ID FROM TABSYSSESSIONLIST LEFT JOIN
LSTNAMEADDRESSCAT ON TABSYSSESSIONLIST.ID=LSTNAMEADDRESSCAT.ID WHERE
SESSION='..' AND LIST='0' ORDER BY LASTNAME, FIRSTNAME;
When it comes as the first SQL statement it takes about 0.35 sec to
execute. First I thought there was something wrong with it, but I
couldn't find anything.
But then I found out that, if I put a very simple SQL statement first
(something like SELECT ID FROM TABLE;) and the one above as the second
one, the one above takes only about 0.05 seconds to execute. But then
the simple first one takes more then 0.3 secs. Anyone has a clue?

Marc

--- In firebird-php@yahoogroups.com, "marcdecaluwe"
<marcdecaluwe@y...> wrote:

>
>
> I'm using Firebird 1.0.0.
> Connection times vary from 0.2 sec to over 2sec.
>
>
> --- In firebird-php@yahoogroups.com, "Ard Biesheuvel"
> <a.k.biesheuvel@e...> wrote:
> >
> > --- In firebird-php@yahoogroups.com, "Alan McDonald" <alan@m...>

wrote:

> >
> > > of some pages. I have been analysing the pages with timers and it
> > > seems that alls SQL command seem to run rather fast, but to my
> > > surprise the time to make a connection to the database
> (ibase_connect)
> > > and the FIRST SQL statement seem to take about 80% of all the
> time the
> > > script needs to run. Even if I change the first SQL statement,
> > > execution time can vary a bit depending on the statement

itself, but

> > > it always takes a LOT more time than when I execute it as a

second.

> > > I experimented also a bit with ibase_pconnect but couldn't

find any

> > > improvement. Maybe I'm missing something somewhere?
> >
> > Are you using the most recent release of Firebird 1 (not 1.5) ?
> > IIRC, at some
> point a bug was fixed that caused connection requests to
> > be delayed for some time for no reason.
> >
> > --
> > Ard