If I'm not mistaken, the new savepoint capability in Firebird is accessed
via an actual SQL statement passed in via the ExecuteImmediate() method. The
statement you pass in would be something like 'SAVEPOINT <some name>'.

This is different than the TIB_Transaction.SavePoint method. It performs a
commit retaining without impacting datasets in an edit state. It allows them
to remain unposted. Calling TIB_Transaction.CommitRetaining does enforce
that all associated datasets post if they are in an edit state.

As for your question/problem. I will probably want to see an example of this
odd behavior.

HTH,
Jason Wharton

----- Original Message -----
From: "james_027" <james_027@...>
To: <IBObjects@yahoogroups.com>
Sent: Tuesday, September 16, 2003 12:38 AM
Subject: Re: [IBO] weird posting error


> Hi
>
> Iam talking about the savepoints in IBO. Where Jason says what it
> does is its commits the works of the current dataset without
> requering the other dataset set to commits it work. I have tested it.
> For example i have ib_query1 and ib_query2 were both of them are in
> insert mode for example and when i have this
> ib_query1.ib_transaction.commitretainning, it gives me a error
> message because the ib_query2 is not save yet and in insert state.
> but if I use the ib_query1.ib_transaction.savepoint it is fine. Then
> I use ibexpert to check the the record was really inserted, yes it
> does.
>
> So what do you think is the possible wierd posting error? because I
> already encounter that twice within the month.
>
> If you suggest me not to use the ib_transaction.savepoint, do you
> have other solution? considering that I have other ib_query in insert
> state or edit state?
>
> thanks a lot
>