From Newsgroup: comp.databases.ingres

Hi,
I've this OR6.2 program that deletes data from an Ingres 11 database on the same Windows 2012 R2 host.
As Ingres 11 and OR6.2 don't (or did not ) play nice (Ingres Net issues as far as I recall) they are on separate installations on the host.
For a very specific multi-row (28 rows) delete that looks like this:
Delete from relevant_date where case_id = 700431
The OpenROAD (and the SQL prompt using the same OR6.2 vnode) causes this error when deleting a record:
E_DM0056_NOPART NOPART Btree error.
E_DM9385_NO_INDEX_TUPLE Cannot locate secondary index row for base table row at TID 15373.
E_DM9386_SEC_INDEX_ERROR Error updating secondary index ,relevant_date_idx, for change to base table row at TID 15373.
E_DM9387_INDEX_UPDATE_ERROR Update failed due to inability to update secondary index table.
The relevant_date_idx is a secondary index on a date column in the relevant_date table.
In the OR6.2 installation the II_TIMEZONE_NAME was set to GMT and in Ingres 11 UNITED-KINGDOM
Setting the OR6.2 to UNITED_KINGDOM makes it all work.
The row in question under GMT gives the index column date as '21/10/1999 00:00:00' but under UNITED-KINGDOM looks like '21/10/1999 01:00:00'.
I settle for working, understanding is a luxury that few of my clients are willing to pay for.
But there is a part of me that thinks - "Is that by design that it looks to compare the key values and the TID/TIDP when deleting from the primary table?"
I figured that this is some kind of adult stuff that us kiddies just donrCOt get.
Thoughts?
Cheers
Adrian
--- Synchronet 3.21b-Linux NewsLink 1.2