From: Sergei Golubchik
Date: March 4 2010 5:54pm
Subject: Re: Change request to WL#3561 - transactional LOCK TABLE
List-Archive: http://lists.mysql.com/internals/37776
Message-Id: <20100304175439.GA24538@janus.mylan>
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit

Hi, Ingo!

On Mar 04, Ingo Strüwing wrote:
> 
> Sinisa requests to change WL#3561 (transactional LOCK TABLE) so that
> READ/WRITE locks are converted to SHARE/EXCLUSIVE locks for
> transactional tables.
> 
> Rationale: Users won't need to change their applications to profit
> from transactional table locks.

How comes ? "Their applications" use UNLOCK TABLES, their applications
access no tables besides explicitly specified in LOCK TABLES, expect
LOCK TABLES to commit an active transaction, expect them to work in
autocommit mode.

I don't see how these applications can profit from transactional table
locks. Even if they don't rely on locks working in autocommit mode, even
if they don't rely on LOCK TABLE committing a transaction, and even if
they don't use UNLOCK TABLE - they certainly, positively, absolutely use
*only* tables that were explicitly locked and *never* lock tables
incrementally. Without changing they will never benefit from
transactional locks no matter what you do in the server.
 
Regards,
Sergei