Re: oracle sequence numbers

From: Pablo Sanchez <pablo_at_dev.null>
Date: Fri, 17 Jan 2003 14:46:58 -0600
Message-ID: <Xns93068C3532BC3pingottpingottbah@216.166.71.233>

rhairgroveNoSpam_at_Pleasebigfoot.com (Bob Hairgrove) wrote in news:3e286478.877301_at_news.webshuttle.ch:

> On Fri, 17 Jan 2003 11:07:58 -0600, Pablo Sanchez <pablo_at_dev.null>
> wrote:
>

>>Ahm, I was looking for a specific DBMS though.  Can you list a
>>commercially available DBMS that has this support?

>
> DB2, for one.
>
> Even lowly MS-Access has a primitive kind of timestamp, but they call
> it DATETIME. If you look at the result set from the ODBC call to
> SQLGetTypeInfo() for Access, you'll see that it has DATETIME with the
> SQL type listed as SQL_TYPE_TIMESTAMP (code 93) with precision of 19,
> min_scale and max_scale both == 0. Also, you can successfully bind
> C-type TIMESTAMP_STRUCT variables to DATETIME columns.
>
> I'm sure much the same is true of Oracle et al.

Somehow this thread has devolved. Let me recap: Someone stated earlier in this thread that "a timestamp instead of a sequence" could be used. I stated that one needs to be careful because timestamps _do_ change.

You said "huh?" and I responded with examples of two DBMS's whose timestamp datatypes can be changed. You additionally stated that "once you have written a value to a column, it should stay the same..." which is clearly incorrect -- as per the examples I gave.

So that portion of the thread is dead and my point remains the same: be careful about using timestamps in place of sequence numbers.

The rest of the thread, although somewhat interesting, is irrelevant to my original point. :)

--
Pablo Sanchez, High-Performance Database Engineering
http://www.hpdbe.com

Received on Fri Jan 17 2003 - 14:46:58 CST