Re: Review: listagg aggregate

Thom Brown <thombrown@gmail.com>

From: Thom Brown <thombrown@gmail.com>
To: Pavel Stehule <pavel.stehule@gmail.com>
Cc: Takahiro Itagaki <itagaki.takahiro@oss.ntt.co.jp>, Robert Haas <robertmhaas@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>, "David E. Wheeler" <david@kineticode.com>, "pgsql-hackers@postgresql.org Hackers" <pgsql-hackers@postgresql.org>
Date: 2010-02-01T13:51:42Z
Lists: pgsql-hackers
On 1 February 2010 13:40, Pavel Stehule <pavel.stehule@gmail.com> wrote:

> 2010/2/1 Thom Brown <thombrown@gmail.com>:
> > I noticed that the regression test results don't include the following
> case:
> >
> > select string_agg(a) from (values(null),(null)) g(a);
> >
> > There is one similar where a delimiter is provided.
> >
> > Which leads me to ask for clarification, would it be safe to assume that
> > string_agg can never itself return null?
>
> if all values are null, then result is null.
>
> Pavel
>
>
>
Ah, I was looking at the expected results, and couldn't see a NULL outcome,
but then these aren't indicated in such results anyway.

Thom