Обсуждение: Client software vs cursor problem?

I've got a client trying to access PG through ODBC from something called ENVOX
(a telecomms IVR system). It lets you build control scripts visually by
dragging/dropping boxes and setting some params.

Now, if he uses a SELECT box to open a (static) cursor then an INSERT box to
add a record it fails, but *only* if the select returns no records. If the
INSERT box is replaced by a "RAWSQL" box he can successfully build his own
insert statement and run it. He's happy enough with this workaround, but it'd
be good if we can figure out what's happening.

The error he gets is "row value out of range" which appears to be HY107 from
googling. This suggests to me the INSERT box is trying to move to the
first/last row or similar and failing because there are no rows.
Unfortunately I'm no ODBC expert and can't see what code underlies the box.

Two questions:
1. Does it sound like I'm on the right track?
2. Would getting him to turn on odbc-logging at his end help me to diagnose?
(he's remote from me).

Oh - we've tried variations on the cursor type and cursor-concurrency settings
and they seem to make no difference.

TIA
--
  Richard Huxton

Which version of psqlodbc driver are you using ?

regards,
Hiroshi Inoue
    http://www.geocities.jp/inocchichichi/psqlodbc/

Richard Huxton wrote:
>
> I've got a client trying to access PG through ODBC from something called ENVOX
> (a telecomms IVR system). It lets you build control scripts visually by
> dragging/dropping boxes and setting some params.
>
> Now, if he uses a SELECT box to open a (static) cursor then an INSERT box to
> add a record it fails, but *only* if the select returns no records. If the
> INSERT box is replaced by a "RAWSQL" box he can successfully build his own
> insert statement and run it. He's happy enough with this workaround, but it'd
> be good if we can figure out what's happening.
>
> The error he gets is "row value out of range" which appears to be HY107 from
> googling. This suggests to me the INSERT box is trying to move to the
> first/last row or similar and failing because there are no rows.
> Unfortunately I'm no ODBC expert and can't see what code underlies the box.
>
> Two questions:
> 1. Does it sound like I'm on the right track?
> 2. Would getting him to turn on odbc-logging at his end help me to diagnose?
> (he's remote from me).
>
> Oh - we've tried variations on the cursor type and cursor-concurrency settings
> and they seem to make no difference.

On Thursday 26 Jun 2003 4:43 am, Hiroshi Inoue wrote:
> Which version of psqlodbc driver are you using ?

Can't get the details until this evening, but very recent - about a month old.

--
  Richard Huxton

Richard Huxton wrote:
>
> On Thursday 26 Jun 2003 4:43 am, Hiroshi Inoue wrote:
> > Which version of psqlodbc driver are you using ?
>
> Can't get the details until this evening, but very recent - about a month old.

There was a fix recently.
Please try the snapshot dll at
    http://www.geocities.jp/inocchichichi/psqlodbc/ .

regards,
Hiroshi Inoue
    http://www.geocities.jp/inocchichichi/psqlodbc/

On Monday 30 Jun 2003 5:14 am, Hiroshi Inoue wrote:
> Richard Huxton wrote:
> > On Thursday 26 Jun 2003 4:43 am, Hiroshi Inoue wrote:
> > > Which version of psqlodbc driver are you using ?
> >
> > Can't get the details until this evening, but very recent - about a month
> > old.
>
> There was a fix recently.
> Please try the snapshot dll at
>     http://www.geocities.jp/inocchichichi/psqlodbc/ .

Thanks Hiroshi, I'll give that a go (probably tomorrow).
--
  Richard Huxton