Re: Torque using Tomcat

[Raphael Mankin <ra...@mankin.org.uk>]

I have exactly the same issue, that the session does not persist over a
restart, except that I am *not* using JNDI.


>From the Tomcat logs I can see that there is a serialisation issue associated
with the application restart. The sessions used to persist but some changes in
the application have caused it not longer to do so. Therefore, for me at least,
I am fairly sure that this is an application issue and not a
Torque/Struts/Tomcat issue.


On 31-Dec-2004 Saravana Krishnan wrote:
> Hi,
> 
>   This is a late reply to this Tomcat problem. But I have no problems
> using Torque 3.1 (or 3.1.1 not sure)  with Tomcat 4.x. The main
> application in which I do development is a jar file when it is started
> in Tomcat. If I make some changes, to the java files, I recompile and
> create a jar file and simply overwrite the jar file that Tomcat is
> using. When I do this, within a span of 10 secs, Tomcat realizes that
> the jar file has changed and restarts itself. Torque is reinitialized
> and everything works without a problem except the fact that the session
> before the restart is no more there. I don't need the session to exist
> between restarts, so it's not an issue for me. I can give more details
> on how I configure this if anyone is interested. Btw, yes I use a JNDI
> datasource that is provided by Tomcat.
> 
> Thanks,
> Sarav
> 
> On Thu, 2004-12-30 at 08:33 +0000, Henning P. Schmiedehausen wrote:
>> "M. Sean Gilligan" <Se...@catalla.com> writes:
>> 
>> Hi,
>> 
>> 
>> >>"M. Sean Gilligan" <Se...@catalla.com> writes:
>> >>
>> >>I took a good look at the JndiDataSourceFactory and this patch just
>> >>plasters over
>> 
>> >What are the consequences of a patch that just "plasters over"?
>> 
>> That one just made sure that the config loop was not run when the
>> object was empty. The deeper problem is, that the loop _should_ be run
>> when it is empty, but should have no effect (which it had, because
>> after the loop, but inside the if (c == null) statement, the
>> datasource was blindingly initialized.
>> 
>> I cleaned the code to make sure that this does not happen. "Plastering
>> over" in my personal speaking means that you simply add "if (xxx ==
>> null)" or "if (foo.isEmpty())" tests to a code logic without solving
>> the inherent problems with the control flow. In this case, it was
>> possible to add a null DataSource object which is a bad thing in
>> itself. And it does not matter, _how_ we got there but that the
>> program logic allows it.
>> 
>> >>(there is also a good NPE chance in there)
>> 
>> >I'm sorry, I'm not familiar with the acronym NPE, what does "NPE" mean?
>> 
>> NullPointerException. :-) 
>> 
>> >>I would prefer having you test the TORQUE_3_1_BRANCH HEAD by doing
>> >>
>> >>cvs -d :pserver:anoncvs@cvs.apache.org:/home/cvspublic checkout -r
>> >>TORQUE_3_1_BRANCH db-torque
>> >>
>> >>and compiling it,
>> 
>> >I did this and it fixed the problem, but...
>> 
>> >> but if you are stuck to a released version due to QA
>> >>reasons, simply take the released 3.1.1 version and replace the
>> >>JndiDataSourceFactory.java file.
>> 
>> >I really do want to use a QA'ed release for my alpha/beta/production
>> >release, so I also tried copying JndiDataSourceFactory.java into the 3.1.1
>> >directory tree and am unable to do a build:
>> 
>> Yep, now that you say it: You will probably need all the contents of
>> the dsfactory package because the inheritance of the DataSource
>> factories changed.
>> 
>> For a released version, I do plan to put out 3.1.2 early next year
>> (which means probably around end of January), depending on how much
>> issues will crop up and what the other developers think.
>> 
>> [...]
>> 
>> >Any suggestions on what to do with the 3.1.1 tree?  I'm going to try the
>> >"plasters over" patch for now...
>> 
>> I didn't mean to use "plaster over" as a derogatory term. I was just
>> stating a fact. Sorry if you felt offended.
>> 
>>      Regards
>>              Henning
>> 
> -- 
> Saravana Krishnan <sa...@uky.edu>
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: torque-user-unsubscribe@db.apache.org
> For additional commands, e-mail: torque-user-help@db.apache.org
> 

-- 
                               Lead us not into temptation; we
                               can find our own way.


Raphael Mankin
----------------------------------

---------------------------------------------------------------------
To unsubscribe, e-mail: torque-user-unsubscribe@db.apache.org
For additional commands, e-mail: torque-user-help@db.apache.org