To: "Marc G. Fournier" <scrappy@postgresql.org>
Cc: pgsql-sql@postgresql.org
Subject: Re: PL/SQL Function: self-contained transaction?
References: <20050822141357.B1002@ganymede.hub.org>
In-Reply-To: <20050822141357.B1002@ganymede.hub.org>
From: Greg Stark <gsstark@mit.edu>
Organization: The Emacs Conspiracy; member since 1992
Date: 22 Aug 2005 14:20:00 -0400
Message-ID: <8764txj0kv.fsf@stark.xeocode.com>
Lines: 25
User-Agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.3
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii


"Marc G. Fournier" <scrappy@postgresql.org> writes:

> In PostgreSQL, as everyone knows, a QUERY == a transaction, unless wrap'd in a
> BEGIN/END explicitly ... how does that work with a function?  is there an
> implicit BEGIN/END around the whole transaction, or each QUERY within the
> function itself?

The whole outer query issued from your frontend is in one transaction.

> If the whole function (and all QUERYs inside of it) are considered one
> transaction, can you do a begin/end within the function itself to 'force'
> commit on a specific part of the function?

Functions cannot issue start or end transactions. They're a creature of the
transaction you're in when you call them. Otherwise it wouldn't make sense to
be able to call them from within a query.

There is some discussion of "stored procedures" which would live outside of
transactions and be able to create transactions, commit, and roll them back.
But I don't think any of that work is committed yet. I'm not even sure it's
been written yet.

-- 
greg