On Tuesday 17 February 2004 23:33, Tom Lane wrote:
> Richard Huxton <dev(at)archonet(dot)com> writes:
> > 2. Write a custom aggregate function (like sum()) to do the
> > concatenation. This is easy to do, but the order your ABC get processed
> > in is undefined.
> Actually, as of 7.4 it is possible to control the order of inputs to a
> custom aggregate. You do something like this:
> SELECT foo, myagg(bar) FROM
> (SELECT foo, bar FROM table ORDER BY foo, baz) AS ss
> GROUP BY foo
> The inner sub-select must order by the columns that the outer will group
> on; it can then order by additional columns that determine the sort
> order within each group. Here, myagg() will see its input ordered by
> increasing values of baz.
Hmm - good to know, but I'm always wary of doing this sort of thing. It's
exactly the sort of trick I look at 18 months later, fail to read my own
comments and "tidy" it.
In response to
pgsql-sql by date
|Next:||From: Richard Huxton||Date: 2004-02-18 09:40:16|
|Subject: Re: CHAR(n) always trims trailing spaces in 7.4|
|Previous:||From: Achilleus Mantzios||Date: 2004-02-18 07:18:16|
|Subject: Re: bytea or blobs?|