Sent using Zoho Mail
---- On Fri, 12 Mar 2021 20:24:48 -0500 Christopher Schultz
<ch...@christopherschultz.net> wrote ----
>
> Alex,
>
> On 3/12/21 16:32, My Subs wrote:
> > ---- On Fri, 12 Mar 2021 02:35:27 -0500 Mark Thomas <ma...@apache.org>
> > wrote ----
> >
> > > On 12/03/2021 03:57, My Subs wrote:
> > > > Hello,
> > > >
> > > > I'm using Tomcat 10.0.0. Suppose I call setAutoCommit(false) on a
> > connection obtained from a Tomcat JDBC Connection Pool. Then I do some
> > stuff with the connection, call commit() or rollback() and finally call
> > close() on it without ever calling setAutocommit(true).
> > > >
> > > > What will the autocommit state of that connection be, the next time
> > the pool gives it to my code? Will it be in the same state I left it,
> > that is, with autocommit set to false, or will it be reverted back to the
> > default state (autocommit set to true)?
> > >
> > > The simplest way to be sure is to run a test with a pool size of 1.
> > >
> > > A quick look at the code (I might have missed something) suggests that
> > > if defaultAutoCommit is configured then it will be reset to that
> > default
> > > value. Otherwise it will be unchanged from when the connection was
> > > returned to the pool.
> > >
> > > Mark
> > >
> > > ---------------------------------------------------------------------
> > > To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org
> > > For additional commands, e-mail: users-h...@tomcat.apache.org
> >
> > Hi Mark,
> >
> > Thanks for your comment. I found this on
> > https://tomcat.apache.org/tomcat-10.0-doc/jdbc-pool.html:
> >
> > "The only state the pool itself inserts are defaultAutoCommit,
> > defaultReadOnly, defaultTransactionIsolation, defaultCatalog if these are
> > set. These 4 properties are only set upon connection creation. Should
> > these properties be modified during the usage of the connection, the pool
> > itself will not reset them."
> >
> > It seems to say that connections will not be reverted back to the default
> > auto-commit state even if defaultAutoCommit is set. Given your reading of
> > the source code, might it be that the docs are wrong?
>
> I think the docs can easily be misunderstood.
>
> I don't use the tomcat-pool but I do use the *other* pool provided by
> Tomcat and they are generally expected to be (mostly) interchangeable.
>
> A connection pool which doesn't reset the auto-commit state on
> connection-return operation would IMO be considered very badly broken.
>
> tomcat-pool allows you to NOT specify a default in which case
> Connection.setAutoCommit() will never be called *at all* by the pool.
> That seems like reasonable behavior (don't mess with my stuff!), but it
> means that the application will *always* have to reset the auto-commit
> state to a known value *every single time* a connection is borrowed.
>
> Since the whole point of the pool is to manage this kind of thing, I
> would argue that having defaultAutoCommit NOT set to anything would be
> considered very bad practice.
>
> If you set defaultAutoCommit="true", you should expect that, when your
> connections are returned to the pool, that setAutoCommit(true) will be
> called every single time the connection is returned to the pool --
> usually by the application calling Connection.close().
>
> Are you worried about a particular use-case or are you just being
> extra-vigilant? Or are you observing some unexpected behavior?
>
> -chris
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org
> For additional commands, e-mail: users-h...@tomcat.apache.org
>
>
Hi Chris,
I’m trying to wrap my head around on whether I can end transaction blocks in a
try clause with a simple call to commit(). That would be the case if the pool
resets the auto-commit state to the value of defaultAutoCommit. Try blocks
would end with commit(), catch blocks with rollback(), and the pool takes care
of setting auto-commit back to true.
On the other hand, if as the docs say, “the pool itself will not reset” the
state of connection attributes like auto-commit, then I’d need to tack to every
try/catch a finally clause having setAutoCommit(true). But this would be
unnecessary boilerplate if, as Mark says —and you believe should be the case—,
the pool does effect the reset.
I’d rather not bloat the code, hence my question. I guess that puts me in the
extra-vigilant camp.
Cheers,
Alex
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org
For additional commands, e-mail: users-h...@tomcat.apache.org
