Archives of the TeradataForum
Message Posted: Thu, 18 Aug 2005 @ 08:17:12 GMT
Subj: | | Re: Deleting Duplicate rows |
|
From: | | Dennis Calkins |
On Wed, 17 Aug 2005 14:31:26 -0400, TeradataForum wrote:
| Here is a way of deleting duplicate rows the way you want it... | |
>3) delete from test1 where substr(rowid,9,1) > '01'xb;
This Type of usage is one of the reasons we now disallow this.
Your assumption here is a HASH Duplicate means you have a Duplicate Row.
What about 2 different values which generate the same HASH value. Called a HASH SYNONYM. You just deleted rows which weren't duplicates.
OOPS.
Also you are are assuming you have < 256 duplicates since the UNIQ[0] will equal 0x00 when we reach 256 hash duplicates and every 256th
value after that.
Bad things happen when you make assumptions about the way something is implemented.
|