Re: default select order
From: Palooka <nobody_at_nowhere.com>
Date: Thu, 08 Jan 2009 23:12:30 +0000
Message-ID: <yLv9l.255107$LM6.242379_at_newsfe03.ams2>
Mark D Powell wrote:
>
> Now I understand what you were referring to. I have to agree though I
> wonder if some of the target buffers already being in the buffer cache
> are affecting the results. I have seen where the key selected from an
> IOT does not return in order. My guess which I never got to test is
> that it was due to blocks already being in the buffer cache and being
> accessed in cached rather than retrieved order.
>
> Without the same full version being available, trying to duplicate the
> problem may be impossible.
>
Yes, but all this is dancing on the head of a pin. Without an order by, Oracle will return the results in whatever order is convenient. We all know that, and IMHO that is all which needed to be advised to OP. As DDF has said, even the OP's subject is an oxymoron; for heap tables there is no "default sort order".
Date: Thu, 08 Jan 2009 23:12:30 +0000
Message-ID: <yLv9l.255107$LM6.242379_at_newsfe03.ams2>
Mark D Powell wrote:
>
> Now I understand what you were referring to. I have to agree though I
> wonder if some of the target buffers already being in the buffer cache
> are affecting the results. I have seen where the key selected from an
> IOT does not return in order. My guess which I never got to test is
> that it was due to blocks already being in the buffer cache and being
> accessed in cached rather than retrieved order.
>
> Without the same full version being available, trying to duplicate the
> problem may be impossible.
>
Yes, but all this is dancing on the head of a pin. Without an order by, Oracle will return the results in whatever order is convenient. We all know that, and IMHO that is all which needed to be advised to OP. As DDF has said, even the OP's subject is an oxymoron; for heap tables there is no "default sort order".
Palooka Received on Thu Jan 08 2009 - 17:12:30 CST