Re: UPDATE ... RETURNING atomicity

=?UTF-8?Q?Grzegorz_Ja=C5=9Bkiewicz?= <gryzman@gmail.com> writes:
> find in docs part that talks about transaction isolation levels, and
> translate it to your problem.

Yes, please read the fine manual:
http://www.postgresql.org/docs/8.4/static/mvcc.html

What I think will happen in your example is that all concurrent
executions will locate the same row-to-be-updated.  The first one to get
to the row "wins" and updates the row.  All the rest will fail, either
updating no rows (if not serializable) or throwing an error (if
serializable).

            regards, tom lane