Subject Re: [IBO] Cursor unknown error -504
Author Bob Zirbel
Thanks for your help Helen and Alan.

I will look at the code again with the keylinks idea in mind.



--- In, Helen Borrie <helebor@t...> wrote:
> At 12:23 AM 9/08/2004 +0000, Bob Zirbel wrote:
> > > Not sure this stacks up Helen, I can leave the app sitting
> >for an hour
> > > and it timer refreshes every 5 minutes. Noone touches it, no
> >errors until
> > > say the 13th or 20th save, then on this 13th or maybe 20th
> >the cursur
> > > unknown error pops up. That's on my dev platform with one user
> >sitting on
> > > it.
> > >
> > > Alan
> >
> >I agree with Alan, Helen, my app is only displaying data in the
> >grid. As far as I know, unless there is a bug, there are no open
> >transactions.
> First, if there is data in the grid, then there is an open
transaction. If
> you are refreshing something by timer, then you are closing and
> datasets. If you are using an Autocommit transaction, you are
> cursors.
> This error is about a database cursor - it's the client asking the
> to work from a set pointer in a cursor that either has gone or
never was
> there.
> Alan's thoughts about DB_KEY would be a useful direction to look,
if your
> table doesn't have a primary key and/or the application doesn't
know what
> the primary key is, for whatever reason. Keylinks that are
present but
> wrong would cause different exceptions, which you haven't reported.
> Under the hood, the application is asking to do something to a row
> CURRENT OF a certain unique key. If the only unique key in your
set, known
> to the client and the server, is the DB_KEY, then the cursor
pointer will
> only be valid if the same record can be found by locating the
db_key that
> the application thinks it has. Remember, the db_key refers to a
> relative location, it's not a data key. It won't be found at that
> if someone (including your user) updated the record and the new
> version has been written to a different relative location; or if
> deleted the record.
> For me, this thread has become quite muddled, with a confused mix
> problem descriptions and conditions, so I'll be happy to leave it
to you
> and others to resolve.
> Helen