--- In Firebird-Java@yahoogroups.com, "phil_hhn" <time_lord@h...> wrote:

>
> --- In Firebird-Java@yahoogroups.com, "Roman Rokytskyy"
> <rrokytskyy@a...> wrote:
> > > This is rather strange because we can run some code when we're not
> > > even touching records that contain blobs. We haven't selected any
> > > records containing blobs.
> >
> > Hmmm... I remember there was similar issue some time ago, but I
> think that
> > bug was already fixed. Are you able to provide a test case that at

least

> > under some conditions reproduces this problem? There it was related
> to the
> > multi-threaded access to the same connection object.
>
> It is good (in a way) that this problem seems familiar! ;-) The
> fbserver version we have is 1.5.0.4306 - could this problem be fixed
> in 1.5.1?
>
> Unfortunately we created a simple database with just the 2 tables
> mentioned, but could not reproduce the problem. So we assume that it's
> with a more complex structure that the JDBC has problems. This is also
> assuming that the data has not got corrupted (assuming there are no
> internal Firebird overruns, etc) or something similar. This problem
> occurs on our system with a freshly created database with freshly
> inserted data.
>
> Although our full-blown database is much more complex than the snippet
> I attached earlier, the code we are running is the same as described,
> i.e all we are doing is updating the VisualSearch table when the error
> occurs.
>
> > If you are not selecting any record with the BLOB fields, then
> something is
> > mixed inside driver, so it thinks that some field is blob and
> interprets its
> > data as BLOB data.
>
> Hmmm.... any way of getting an internal trace of what the driver is
> doing that may show if this is happening?

Well, upgrading to server 1.5.1 didn't help (I was hoping that wasn't
the reason ;-) ).

It suddenly occurred to me that this Win2k problem could be similar to
the other problem we had recently (where calling .next() gives the
error "SQL error code = -504, Cursor"). So we went back through some
old drivers - we have found that with JDBC driver 1.5.1 this error
does not occur. But with 1.5.2 it does.
(I've updated that other post also.)

So what is the change between 1.5.1 and 1.5.2 that breaks this under
win2000 only?