Problem with postgres 9.6 ODBC drivers

From: "Pandian, Jegan" <Jegan(dot)Pandian(at)dell(dot)com>
To: "pgsql-odbc(at)postgresql(dot)org" <pgsql-odbc(at)postgresql(dot)org>
Subject: Problem with postgres 9.6 ODBC drivers
Date: 2017-04-11 07:49:16
Message-ID: 6591B4AAC7D5944EBA4E72533AAE274E4F3F21AC@MX301CL01.corp.emc.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-odbc

Hi Postgres ODBC team,

Our application is failing with the latest windows odbc (postgres) drivers, because it's not updating sequence table at al.
So we will always get the same id when we try to create any new object, ending up in unique constraint violation error.
Debugged postgres odbc driver code (https://www.postgresql.org/ftp/odbc/versions/src/) and found the below:

File: results.c
Method: static void KeySetSet(const TupleField *tuple, int num_fields, int num_key_fields, KeySet *keyset)
Problematic change: keyset->status = 0;

We will execute all queries with status 1024, now for some reason status has been re-initialized to 0. Because of this, update call itself is not getting invoked (code below)

spos_callback function:
#define CURS_IN_ROWSET (1L << 10)
if (0 == (res->keyset[kres_ridx].status & CURS_IN_ROWSET))
continue;

status: 34
CURS_IN_ROWSET: 1024
34 & 1024 => 0

If the status was not reset, then keyset[kres-ridex].status will be 1024+ 34 => 1058 (behavior for 9.4 and 9.5 drivers), so will do the update properly.
We need your help to find out why this has been changed in the latest drivers? Is there any specific reason behind this? Is this a bug/any workaround to overcome it?

Thanks,
Jegan

Browse pgsql-odbc by date

  From Date Subject
Next Message Jonathan Byrd 2017-04-12 17:04:12 psqlODBC on OSX problems
Previous Message Mateusz Loskot 2017-03-28 14:40:26 Re: COLUMN_SIZE for bytea now equals -4 (09.05 vs 09.06)