First of all, thanks for your assistance.

--- In IBObjects@yahoogroups.com, Helen Borrie <helebor@t...> wrote:
> OK; what I meant was, would you describe the transaction setup. So
>
> 1) Are you
> a) using the internal transaction as the DefaultTransaction of the
> ib_connection or
> b) are you assigning an explicit ib_transaction to this property?
>
We've assigned an explicit ib_transaction as the DefaultTransaction.
But that's only there to detect if a programmer uses the connection
without a separate explicit transaction. There's a BeforeStart event
that throws an exception. But again, this transaction never gets used
because the code path is never executed to produce this problem.

(One thing I'm trying to get to here is why you
> would set a ReadOnly transaction to Autocommit...thus causing the
ReadOnly
> transaction to block the advance of the OAT and OIT.)
>
Ignorance :-) I didn't know that doing that would block the OAT.

> Theoretically, you should not need to change the default setting of
> ParameterOrder at all *except* when you have OldParameterOrdering
set to
> 1; in which case it should be set to poOld.
> Logically, if it
> isn't poOld then it is poNew and one should ignore it for Fb 1.5.x
and set
> it to poOld for Fb 1.0.x.
I disagree :-) Someone might want to set OldParameterOrdering=1 on
the FB server so that it can work properly with old software that uses
IBO versions older than 4.3Aa. Then for any new software, using IBO
versions 4.3Aa and greater, leave the default setting to poAuto so
that the new software can detect whatever parameter ordering the
server is using. This seems to me to provide the most flexibility.

>
> > > e. Client library: Brand (Firebird/IB) and build number FB
6.2.3.972
> > (FB 1.0.3) or FB 6.3.3.4842 (FB 1.5.3)
>
> This is a slightly confused mix for addressing the OldParameterOrdering
> issue. With the Fb 1.0.x server and client, only poOld is applicable
> (which the TIB_Connection should detect and set automatically.) The
> correction to the parameter ordering was not present in Fb 1.0.x.
Yes, but I believe the problems is not in FB. I believe the problem
is the IBO Ordering detection logic so it doesn't matter what version
of FB is used.

>
> You are using the unreleased 1.5.3 server and client and switching
back and
> forth between this and Fb 1.0.x. So it seems impossible, on present
> accounts, to determine whether the problem you are seeing arises from a
> regression in the release candidate code ( we don't see any pattern
for Fb
> 1.5.2...) or is due to using an executable that was compiled for one Fb
> version and is being used with another.
I not mixing components of FB versions. I have a clean XP image that
I copy over and then do a clean install of whatever FB version I am
testing.