Re: BUG #5469: regexp_matches() has poor behaviour and more poor documentation

Robert Haas wrote:
> On Mon, May 31, 2010 at 10:11 AM, Bruce Momjian <bruce(at)momjian(dot)us> wrote:
> > Robert Haas wrote:
> >> On Sat, May 29, 2010 at 5:00 PM, Bruce Momjian <bruce(at)momjian(dot)us> wrote:
> >> > I have updated the patch, attached, to clarify that this returns text
> >> > arrays, and that you can force it to always return one row using
> >> > COALESCE() and a '|' pattern (the later suggested by Daniele Varrazzo).
> >>
> >> I don't find this part to be something we should include in the
> >> documentation. ?If we want to include a workaround, how about defining
> >> a non-SRF that just calls the SRF and returns the first row?
> >
> > Remember this has to return one row for no matches, so a simple SRF will
> > not work. ?I also have not seen enough demand for another function. ?A
> > single doc mention seemed the appropriate level of detail for this.
>
> Well, we can debate later whether to add another function to core, but
> what I meant was that the user having the problem could create a
> user-defined function that calls regexp_matches() and returns the
> first row, or NULL.
>
> But actually here's an even simpler workaround, which is IMHO less
> ugly than the original one:
>
> SELECT foo, bar, (SELECT regexp_matches(bar, pattern)) FROM table;

Good idea. Simplified patch attached.

--
Bruce Momjian <bruce(at)momjian(dot)us> http://momjian.us
EnterpriseDB http://enterprisedb.com

+ None of us is going to be here forever. +