 |
 |
Archives of the TeradataForum
Message Posted: Wed, 28 May 2008 @ 11:55:10 GMT
| Subj: | | Re: Dieter's collect stats query on v12 |
| |
| From: | | Anomy Anom |
<-- Anonymously Posted: Wed, 28 May 2008 06:44 -->
Thanks Dieter. Here�s a status. I�ll keep plugging away. I am eliminating the calcs that cause the hiccup, 1 by 1.
I have responded to your thoughts below.
| | Those numeric overflow shouldn't occur if the PadBytes are modified correctly. | |
| | I thought i fixed that, it was due to some unusual default formats. | |
I replaced the padbyte with a 4 - which eliminates the 3610; but introduced the numeric overflow.
| | I remember a 3610 on a 32-bit TD12 beta once, but it seemed to be fixed. | |
| | The strange thing about it was: It only happened when submitted using ODBC, but not with BTEQ. And it was only for the view version, not
the base table version, the only difference between both is an additional calculation in the base table version (but only the view version
crashed). | |
The view version is the culprit. Applications using the algorithm will not be permitted to use the table version.
| | Please try both versions and use SQL Assistant and BTEQ. | |
1) The original incident involved a programmer using BTEQ.
2) It crashed for me using SQL Asst.
| | But still open an incident, because a 3610 is always a bug. | |
Incident opened.
Feedback to follow.
| |
