Re: Multiple return 'columns' from postgre pl/pgsql

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Steve Tucknott <steve(at)retsol(dot)co(dot)uk>
Cc: PostGre <pgsql-novice(at)postgresql(dot)org>
Subject: Re: Multiple return 'columns' from postgre pl/pgsql
Date: 2004-08-09 14:23:11
Message-ID: 14372.1092061391@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-novice

Steve Tucknott <steve(at)retsol(dot)co(dot)uk> writes:
> Does the 'rowtype' have to exist as a definition in the database?

In the form I showed, yes.

> Would returning a record type work -

Only if you're prepared to specify the actual record type in the calling
query. The point is that in

select * from myfunc(...);

the parser has to have some way of understanding what * expands to,
and it needs the info in advance of calling the function. So you
either need to return a named rowtype, or return record and specify
what you're expecting in the call. From memory it's something like

select * from myfunc(...) AS (f1 int, f2 text, ...);

but see the docs. In practice I think the named rowtype is easier in
99% of cases. The returns-record case is really meant for functions
that can actually return different rowtypes depending on the parameters
they are given, like dblink() does. If you're thinking of doing something
like that, you probably shouldn't be asking about it on the novice list ;-)

regards, tom lane

In response to

Responses

Browse pgsql-novice by date

  From Date Subject
Next Message Sean Davis 2004-08-09 14:34:35 Re: Design question regarding arrays
Previous Message Stephan Szabo 2004-08-09 14:22:11 Re: Design question regarding arrays