Re: lock - feedback

From: "Merlin Moncure" <merlin(dot)moncure(at)rcsonline(dot)com>
To: "Thomas Rokohl" <rokohl(at)raygina(dot)de>
Cc: <pgsql-odbc(at)postgresql(dot)org>
Subject: Re: lock - feedback
Date: 2005-10-12 14:18:32
Message-ID: 6EE64EF3AB31D5448D0007DD34EEB3417DD591@Herge.rcsinc.local
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-odbc

> > Have you looked at contrib\userlock? With it, you can simulate
> > pessimistic locks with a non-blocking result code.
> >
> > Merlin
> >
> >
> >
> contrib/userlock? where i can find something about this? in the manual
i

http://developer.postgresql.org/cvsweb.cgi/~checkout~/pgsql/contrib/user
lock/README.user_locks?rev=1.3;content-type=text%2Fplain

> can't find something ?!?!
> mhm what is with the "lost update" problem, can i solve it with it?!?!

probably. However, in many cases proper use of transactions is more
appropriate. This will be even easier to do when we get proper
assertions.

> i can't visualize how it can works?
select user_write_lock(oid) from my_table where value = x;
returns 1 on success, 0 on failure.

just be careful...
select user_write_lock(oid) from my_table where value = x order by y
limit 1;
can acquire more locks than you might think since the table has to be
materialized to do the order.

better to write:

select user_write_lock(oid) from
(
select oid, * from my_table where value = x order by y limit 1;
)

also, don't use oid :). In my applications, I make a domain type called
'cuid' which pulls nextval() from a public sequence. Put that into your
tables and lock on it. Just watch for dump/restore.

Merlin

Responses

Browse pgsql-odbc by date

  From Date Subject
Next Message Thomas Rokohl 2005-10-12 14:51:15 Re: lock - feedback
Previous Message Richard Huxton 2005-10-12 13:58:14 Re: lock - feedback