From: Tom Lane <tgl@sss.pgh.pa.us>
To: "Voillequin, Jean-Marc" <Jean-Marc.Voillequin@moodys.com>
cc: "pgsql-sql@lists.postgresql.org" <pgsql-sql@lists.postgresql.org>
Subject: Re: parallel safe on user defined functions
In-reply-to: 
 <DM6PR20MB3329006C838039F0B3F43E64BEF30@DM6PR20MB3329.namprd20.prod.outlook.com>
References: 
 <DM6PR20MB3329006C838039F0B3F43E64BEF30@DM6PR20MB3329.namprd20.prod.outlook.com>
Comments: In-reply-to "Voillequin,
 Jean-Marc" <Jean-Marc.Voillequin@moodys.com>
	message dated "Wed, 02 Dec 2020 15:25:37 +0000"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <1574256.1606925735.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Wed, 02 Dec 2020 11:15:35 -0500
Message-ID: <1574257.1606925735@sss.pgh.pa.us>
Precedence: bulk

"Voillequin, Jean-Marc" <Jean-Marc.Voillequin@moodys.com> writes:
> For instance, is the following plpython3u function parallel safe?

> create or replace function f1(key text, val_default in text default null=
) returns text as $$
> return GD.get(key.lower(),val_default)
> $$ language plpython3u;

No.  There's no mechanism for sharing Python interpreter state across
processes.

> =3D> Is it a transaction state change?

Throwing an error is always OK (we could hardly decree otherwise).

>                 execute format('select (''%s''::date + interval ''%s''):=
:date',d,concat(n,' day')) into ret;

> =3D> Is this execute statement be considered as a cursor?

No, it's just a weird way of spelling a variable assignment.

			regards, tom lane