I have just deployed an application which uses ADO / ODBC / PostgreSQL (ODBC
driver 7.02.00.02). Although it tested out fine in the lab, in the real
world queries are occasionally returning a spurious last row in the
recordset (after the genuine last row in the table) which contains Null in
every column position. The circumstances in which this occurs are highly
variable - depending on which rows and columns have been selected in the
query. Of course, code which has every right to expect non-null fields
crashes and burns when this happens.
Any suggestions on how to track this down would be appreciated. We have
limited contact with the user group, so it's going to have to be something
application-side that doesn't cause a massive performance hit.
J