On Jul 9, 3:40 pm, "Dave Hughes" <d...@[EMAIL PROTECTED]
> wrote:
> jefftyzzer wrote:
> > Friends:
>
> > In v8 or v9 LUW, if I were to write (in Java or C) my own user-defined
> > aggregate function, would I be able to use it in an OLAP windowing
> > clause as I could use, e.g., COUNT or SUM?
>
> > Example:
>
> > SELECT
> > JEFFS_UDAF() OVER(PARTITION BY SOME_COL)
> > FROM
> > SOME_TABLE;
>
> There's no such thing as a user-defined aggregate function in DB2*,
> external or otherwise, so no you couldn't use it with OLAP windowing
> because it couldn't exist ;-)
>
> Cheers,
>
> Dave.
>
> * Officially at least ... see Knut's fascinating dW article on a
> possible implementation at:
>
> http://www.ibm.com/developerworks/db2/library/techarticle/0309stolze/030
> 9stolze.html
>
> Given that this relies on an existing aggregate function (MAX), I
> suspect it /might/ work with OLAP windowing (though this depends on the
> order of execution).
Hi, Dave:
It was with knowledge of Knut's article that I asked the question. I
suppose another way to rephrase it is: If DB2 sees any ranking/
numbering/aggregation function other than the ones that come "out of
the box" (e.g., one that's user-defined) before the "(OVER PARTITION
BY...)" clause, will it throw an SQL0109N error?
Regards,
--Jeff
|
