proxool-developer — Developer's discussion list

[Proxool-developer] Proxool and JNDI

[<ca...@ab...>]

Hi,

  I want to store a proxool database pool within a JNDI tree, I have seen that
there are some properties to configure it, but I can´t make it works. 

  How should this be done?.

Thank you!!

Re: [Proxool-developer] Re: Error due to Interruption of Thread

[Jean-Henry B. <ber...@ne...>]

bi...@lo... wrote:

>  
>
>>The thread does NOT get interrupted.
>>We are using Proxool.0.8.2
>>
>>Could you provide any suggestions? How would I interrupt the Thread
>>without this error?
>>    
>>
>
>Not sure. Nobody has ever asked to do that before. What sort of behaviour
>would you expect?
>
>I suppose that if we spot the difference between InterruptedExceptions
>that occur in the lock code and those originating from elsewhere we can
>just throw the InterruptedException when approriate.
>
>- Bill
>  
>
I guess Throwable.getCause() can be very handy in a case like this.
(make sure the lock code initializes the "cause" and when checking, if
the cause is different, even if it was not initialized - i.e. null - it 
would mean
it's not generated by the locking code).
Alas all this depends on the platform version , since the "cause"
handling is only available since version 1.4.

Jean

[Proxool-developer] Re: Error due to Interruption of Thread

[Bill H. <bi...@lo...>]

Parambir,

I'm copying your email to the dev list. You should send all questions to
one of the lists so that a) you get more people thinking about your
problem, and b) other people can learn from the solution.

On Thu, 26 Feb 2004, Parambir Gill wrote:

> I am running a process in a Thread. This process accesses the DB
> through Proxool. If i call Thread.interrupt() on this thread, I
> get the following error:
>
> 1241365 [Thread-9] ERROR proxool.AbstractProxyConnection  - Unable
> to acquire write lock for status

Our concurrent locking code (from Doug Lea) can throw an
InterruptedException and we are incorrectly assuming that this exception
originates from that code. That is it a bug.

> The thread does NOT get interrupted.
> We are using Proxool.0.8.2
>
> Could you provide any suggestions? How would I interrupt the Thread
> without this error?

Not sure. Nobody has ever asked to do that before. What sort of behaviour
would you expect?

I suppose that if we spot the difference between InterruptedExceptions
that occur in the lock code and those originating from elsewhere we can
just throw the InterruptedException when approriate.

- Bill

[Proxool-developer] Re: [Proxool-user] java.security.AccessControlException?

[Bill H. <bi...@lo...>]

Hi Kerry.

On Sat, 2004-02-21 at 01:54, Kerry Hew wrote:

> I just came across Proxool and am trying to test it out.  I am able to
> get it up and running fine locally, but when I run the same stuff
> (diff't prop file for diff't db) on my web host, I get the following
> exception seen at the bottom of this message.

Can you tell us where you have installed the proxool.jar file (for
example, in WEB-INF/lib)? Does your application work in other respects? 

> But does it perform?  

Yes, I believe it does. We are running it on some high usage web sites
without problem. 

> What are alternatives to compare to?

Poolman, C3PO, DBCP. Anybody think of any more?

- Bill

Re: [Proxool-developer] Two errors in killConnecton

[Bill H. <bi...@lo...>]

Hi Dorte,

On Tue, 2004-02-10 at 10:56, Dorte Munk-Jakobsen wrote:

> oracle implementation of enqueue and dequeue on the session queues
> leaves some cursors open and after some time there is no more cursors
> left on the connection.

Ugh.

> try {
> 
> proxyConnection = (ProxyConnection)
> proxyConnections.get(nextAvailableConnection);
> 
> } catch (IndexOutOfBoundsException e) {
[snip]
> }

Thanks for that. I'll commit that fix. 

> The second is a little more subtile as you can see below:
> 
> killConnecton(String alias, long id, boolean merciful)
> 
> uses
> 
> ConnectionPoolManager.getInstance().getConnectionPool(alias).expireConnection(id, merciful);
> 
> to call:
> 
> expireConnection(long id, boolean forceExpiry)
> 
> The documentation says merciful=false => we force and the value false
> is passed to expireConnection.forceExpiry where false => don't force
> so ...
> 
> it is easy to make a work around but I suggest that you negate
> merciful when you pass it to forceExpiry.

Also done. Thanks again. 

> I wonder if you know a good site or a good book where I can learn some
> more about proxies? 

I'm not sure I can think of a simple answer to that question. Cglib
might be a good resource if you want to deep into it - that's the code
we use to implement proxies.

- Bill

Re: [Proxool-developer] ThreadDeath Error

[Bill H. <bi...@lo...>]

Hi Shaun,

On Mon, 2004-02-09 at 23:45, Shaun Stepanowich wrote:

> SEVERE: Housekeeping log.error( :
> java.lang.ThreadDeath
>         at
> org.apache.catalina.loader.WebappClassLoader.loadClass(WebappClassLoader.java:1252)
>         at
> org.apache.catalina.loader.WebappClassLoader.loadClass(WebappClassLoader.java:1212)
>         at
> java.lang.ClassLoader.loadClassInternal(ClassLoader.java:302)
>         at
> com.enertiasoft.util.pooling.ConnectionPoolStateListener.upStateChanged(Unknown Source)
>         at
> org.logicalcobwebs.proxool.CompositeStateListener.upStateChanged(CompositeStateListener.java:40)

I agree with Christian. The only direct control over that Thread by
Proxool is to cancel it gracefully. That is, after it has completed its
sweep. It seems that your Thread is dying half way through a sweep. Does
it die at different places each time?

It's a daemon thread, so it will die if the JVM dies. You say you get "a
bunch" - can you be more precise? Every 30 seconds?

- Bill

Re: [Proxool-developer] ThreadDeath Error

[Christian N. <chr...@em...>]

My first hunch would be that this has something to do with Proxool shutdown and hot-deploys.

Are you hot-deploying your app?

If you are, are you sure that you shut down the previous Proxool instance properly?

If you use the ServletConfigurator this will be done automatically for you, otherwise you need to call 
ProxoolFacade.shutdown(0) when your app is unloaded (e.g. from a servlets destroy() method.)

Could this be the problem? Please get back to me if its not and I'll investigate further.

CHR

----- Original Message -----
From: "Shaun Stepanowich" 
Date: Mon, 9 Feb 2004 16:45:47 -0700
To: 
Subject: [Proxool-developer] ThreadDeath Error


Hi,
Could i please get some information on why I might be getting this error.  Some common reason etc.  I am having trouble locating the cause.  Thanks a bunch


SEVERE: Housekeeping log.error( :
java.lang.ThreadDeath
        at org.apache.catalina.loader.WebappClassLoader.loadClass(WebappClassLoader.java:1252)
        at org.apache.catalina.loader.WebappClassLoader.loadClass(WebappClassLoader.java:1212)
        at java.lang.ClassLoader.loadClassInternal(ClassLoader.java:302)
        at com.enertiasoft.util.pooling.ConnectionPoolStateListener.upStateChanged(Unknown Source)
        at org.logicalcobwebs.proxool.CompositeStateListener.upStateChanged(CompositeStateListener.java:40)
        at org.logicalcobwebs.proxool.ConnectionPool.setUpState(ConnectionPool.java:816)
        at org.logicalcobwebs.proxool.HouseKeeper.calculateUpState(HouseKeeper.java:234)
        at org.logicalcobwebs.proxool.HouseKeeper.sweep(HouseKeeper.java:155)
        at org.logicalcobwebs.proxool.HouseKeeperThread.run(HouseKeeperThread.java:39)



Shaun Stepanowich, Development Team
 
Enertiasoft, a Premier Technologies Company
150s, 3030 3rd AVE NE Calgary, AB, Canada  T2A6T7
toll free 1 866 ENERTIA (363 7842)
office.   403 228 9292
fax.      403 265 8993
-- 
___________________________________________________________
Sign-up for Ads Free at Mail.com
http://promo.mail.com/adsfreejump.htm

[Proxool-developer] Two errors in killConnecton

[Dorte Munk-J. <dor...@ya...>]

Still working with oracle AQSessions, I am obliged to close my physical connections at irregular intervals, so I cannot use a timeout. The reason I need to close is that the oracle implementation of enqueue and dequeue on the session queues leaves some cursors open and after some time there is no more cursors left on the connection. Since I don't want to wait for the "out of cursors exception", it occurs at 1000 cursors, which I consider a waste of resources, I have created a counter in my connection pool class (the one that calls the real connection pool, currently proxool, before it was oracle but it lost connections) and when a connection has been reused 100 times I close it with ProxoolFacade.killConnecton(pool-alias,conn-id,false). The purpose it to make it close the connection with the mercifull=false => force=true unfortunately there are two bugs here:
 
the first is in proxool.ConnectionPool.expireConnection(long id, boolean forceExpiry)
where the catch catches ArrayIndexOutOfBoundsException, in JDK 1.3.1
the thrown exception is IndexOutOfBoundsException as shown below.

try {

proxyConnection = (ProxyConnection) proxyConnections.get(nextAvailableConnection);

} catch (IndexOutOfBoundsException e) {

nextAvailableConnection = 0;

proxyConnection = (ProxyConnection) proxyConnections.get(nextAvailableConnection);

}

The second is a little more subtile as you can see below:

killConnecton(String alias, long id, boolean merciful)

uses

ConnectionPoolManager.getInstance().getConnectionPool(alias).expireConnection(id, merciful);

to call:

expireConnection(long id, boolean forceExpiry)

The documentation says merciful=false => we force and the value false is passed to expireConnection.forceExpiry where false => don't force so ...

it is easy to make a work around but I suggest that you negate merciful when you pass it to forceExpiry.

Having worked a lot with your source the last week in order to try to implement support for the AQ Sessions (I was successfull, but it did not solve the open cursors problem so I dropped it again) I wonder if you know a good site or a good book where I can learn some more about proxies? 

Thanks

Dorte



---------------------------------
Do you Yahoo!?
Yahoo! Finance: Get your refund fast by filing online

[Proxool-developer] ThreadDeath Error

[Shaun S. <sh...@en...>]

Hi,
Could i please get some information on why I might be getting this
error.  Some common reason etc.  I am having trouble locating the cause.
Thanks a bunch
 
 
SEVERE: Housekeeping log.error( :
java.lang.ThreadDeath
        at
org.apache.catalina.loader.WebappClassLoader.loadClass(WebappClassLoader
.java:1252)
        at
org.apache.catalina.loader.WebappClassLoader.loadClass(WebappClassLoader
.java:1212)
        at java.lang.ClassLoader.loadClassInternal(ClassLoader.java:302)
        at
com.enertiasoft.util.pooling.ConnectionPoolStateListener.upStateChanged(
Unknown Source)
        at
org.logicalcobwebs.proxool.CompositeStateListener.upStateChanged(Composi
teStateListener.java:40)
        at
org.logicalcobwebs.proxool.ConnectionPool.setUpState(ConnectionPool.java
:816)
        at
org.logicalcobwebs.proxool.HouseKeeper.calculateUpState(HouseKeeper.java
:234)
        at
org.logicalcobwebs.proxool.HouseKeeper.sweep(HouseKeeper.java:155)
        at
org.logicalcobwebs.proxool.HouseKeeperThread.run(HouseKeeperThread.java:
39)

 
 

Shaun Stepanowich, Development Team

 

Enertiasoft <http://www.enertiasoft.com/> , a Premier Technologies
Company

150s, 3030 3rd AVE NE Calgary, AB, Canada  T2A6T7

toll free 1 866 ENERTIA (363 7842)

office.   403 228 9292

fax.      403 265 8993

 

RE: [Proxool-developer] Connections closed and open

[Bill H. <bi...@lo...>]

Hi John,

On Thu, 2004-02-05 at 17:20, Hume, John - NA US HQ Delray wrote:

> If the instance of the Connection-proxy is NOT the same one, then this
> problem goes away: calling close() returns the underlying Connection to the
> pool and can set the state of the Connection proxy so that any subsequent
> attempt to use it can throw "SQLException: connection already closed" or
> something like that.

It *is* the same instance of the proxy connection. So this is a problem
I believe. I quite like the idea of wrapping up the connection inside a
disposable wrapper. That would certainly solve this issue.

It's not a quick fix however. There isn't a clean separation from the
wrapper and the internal component that knows about the state of the
connection. It would be good to clean it up though. At the same time, we
could get rid of the redundant abstraction we needed when dealing with
multiple JDKs (back before we used Cglib).

We certainly wouldn't do this before 0.10 (0.9 is due soon to add
DataSource implementation). But maybe it should be done before 1.0.

- Bill

RE: [Proxool-developer] Connections closed and open

[Hume, J. - NA US HQ D. <joh...@cr...>]

Forgive my lack of familiarity with Proxool's architecture, but we've had
some issues with closing connections multiple times (with other pools as
well as proxool) and I want to understand where proxool sits in respect to
this issue currently.

Thread 1 gets a connection from the pool ... what it has is a
Connection-implementing proxy for the underlying driver-supplied Connection.
Thread 1 uses the Connection-proxy just like a normal Connection then calls
close() on it. 

Thread 2 gets a connection from the pool, and let's say it's the same
underlying Connection Thread 1 just returned to the pool.  Might it also be
the same instance of the Connection-proxy?  If so, that's almost surely
trouble, as Thread 1 might still have a reference to it and call close() a
second time, thereby returning it to the pool and allowing some other thread
to get it while Thread 2 is still using it.

If the instance of the Connection-proxy is NOT the same one, then this
problem goes away: calling close() returns the underlying Connection to the
pool and can set the state of the Connection proxy so that any subsequent
attempt to use it can throw "SQLException: connection already closed" or
something like that.

-john.



-----Original Message-----
From: Bill Horsman [mailto:bi...@lo...]
Sent: Thursday, February 05, 2004 11:19 AM
To: Proxool
Subject: [Proxool-developer] Connections closed and open


Hi all,

Just investigating what happens when you close a connection twice. At
the moment, nothing bad happens but you get a nasty warning message
like:

ConnectionPool.java:356) - Unable to set status of connection 40 from
ACTIVE to AVAILABLE. It remains AVAILABLE

We should obviously spot that. I'll have to check but I believe that is
okay to close() twice and if you do then it should do so silently.

It also got me thinking. I don't think we check the connection status
when using it. So it is quite possible to close the connection (i.e.
return it to the pool) and then keep on using it. This would be unwise.
Perhaps we should be throwing an exception if you try to do anything to
a connection that has been closed.

But then, what happens if someone else grabs that connection? Then two
"users" could be using the same one. I'm not sure how to avoid that
problem. Perhaps that is something that the user has to avoid.

Our goal is to behave exactly as if you were using the driver directly
without a pool. And in that scenario the driver would spot it happening
and throw an exception.

I can't think of a solution to this. Maybe it's just something to
document.

- Bill



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
Proxool-developer mailing list
Pro...@li...
https://lists.sourceforge.net/lists/listinfo/proxool-developer

[Proxool-developer] Connections closed and open

[Bill H. <bi...@lo...>]

Hi all,

Just investigating what happens when you close a connection twice. At
the moment, nothing bad happens but you get a nasty warning message
like:

ConnectionPool.java:356) - Unable to set status of connection 40 from
ACTIVE to AVAILABLE. It remains AVAILABLE

We should obviously spot that. I'll have to check but I believe that is
okay to close() twice and if you do then it should do so silently.

It also got me thinking. I don't think we check the connection status
when using it. So it is quite possible to close the connection (i.e.
return it to the pool) and then keep on using it. This would be unwise.
Perhaps we should be throwing an exception if you try to do anything to
a connection that has been closed.

But then, what happens if someone else grabs that connection? Then two
"users" could be using the same one. I'm not sure how to avoid that
problem. Perhaps that is something that the user has to avoid.

Our goal is to behave exactly as if you were using the driver directly
without a pool. And in that scenario the driver would spot it happening
and throw an exception.

I can't think of a solution to this. Maybe it's just something to
document.

- Bill

[Proxool-developer] Re: [Proxool-user] warning messages

[Bill H. <bi...@lo...>]

Hi Dave,

On Tue, 2004-02-03 at 21:28, dav...@fe... wrote:

> We've been using Proxool in production for over a year now. 

That's good. Is that a public web site?

> I just started noticing this error message in our logs:
> 2592 [http8080-Processor25] WARN proxool.pwdds  - 000017 (01/03/00) -
> #3 registered a statement as closed which wasn't known to be open.

That could happen if you close a statement twice. If that is the case
then the only consequence is a WARN log event, but I guess it would be a
good idea not to close it twice. Is it possible that is the problem?

> Also, I just upgraded from 7.x to 8.3 and Im getting these error
> messages now after 30 seconds or so of inactivity:
> 
> 307112 [Shutdown Hook] ERROR proxool.ShutdownHook  - Problem calling 
> "get cause" on IllegalStateException.

That message appears during shutdown. We need to make it a less alarming
debug message.

> org.logicalcobwebs.proxool.ShutdownHook.run(ShutdownHook.java:99)
>         at java.lang.Thread.run(Thread.java:536)

But this line suggests to me (quite strongly) that the JVM is shutting
down. The only way that I can think of for the ShutdownHook.run method
to be called is by the JVM.

What JVM are you using? Do you have any other evidence that might
suggest this is happening?

Do you see log events for Proxool restarting after this shutdown
activity?

- Bill

[Proxool-developer] Hibernate test cases & JIRA

[Bertrand R. <ber...@mo...>]

Dear developpers/maintainers,

I was wondering were I should report issues regarding the Hibernate test
cases...
Should I post a mail to the dev mailing list or should I post an entry in
JIRA ?

If the JIRA solution is the prefered approach, it might be a good idea to
add a 'TestCase' component to the list.

Feedback ?


-bertrand

Re: [Proxool-developer] proxool integrating in tomcat & jmx

[Christian N. <chr...@em...>]

> Hi everybody,
> I just looked at proxool, an it seems very interesting 
> to me since you are able to monitor your connection pool 
> more detailed than you are e.g with the jakarta commons pool, 
> which you use by default in tomcat.
> But there is also my question: 
> Am I able to integrate proxool in my tomcat container by 
> inserting the neccessary configuration details in server.xml ?? 
> It would be good, not to use a application-dependent configuration. 
> And I also would prefer referencing the pool in my application 
> by using a JNDI lookup.

The next Proxool release (0.9 - out in a few weeks time) will add DataSource and ObjectFactory support. So if you have time to wait for that release the answer is yes.

> Next question is, why not building a MBean, 
> which has the same information as the AdminServlet. 
> Then you would be able to extract this data more easily to 
> external systems, like a management appilcation by using a 
> jmx-to-snmp adaptor for instance. As I can see so far you 
> only get information (from the MBean) about non-changing 
> values like the maximum-connection-count not the effectively used connections.

This is a very good suggestion. I'll try to make sure this is included in the Proxool 1.0 release (hopefully out before summer).

CHR

> 
> Thanks for any response,
> Philipp.


-- 
___________________________________________________________
Sign-up for Ads Free at Mail.com
http://promo.mail.com/adsfreejump.htm

[Proxool-developer] proxool integrating in tomcat & jmx

[<phi...@we...>]

Hi everybody,
I just looked at proxool, an it seems very interesting to me since you are able to monitor your connection pool more detailed than you are e.g with the jakarta commons pool, which you use by default in tomcat.
But there is also my question: Am I able to integrate proxool in my tomcat container by inserting the neccessary configuration details in server.xml ?? It would be good, not to use a application-dependent configuration. And I also would prefer referencing the pool in my application by using a JNDI lookup.
Next question is, why not building a MBean, which has the same information as the AdminServlet. Then you would be able to extract this data more easily to external systems, like a management appilcation by using a jmx-to-snmp adaptor for instance. As I can see so far you only get information (from the MBean) about non-changing values like the maximum-connection-count not the effectively used connections.

Thanks for any response,
Philipp.
  
______________________________________________________________________________
Erdbeben im Iran: Zehntausende Kinder brauchen Hilfe. UNICEF hilft den
Kindern - helfen Sie mit! https://www.unicef.de/spe/spe_03.php

Re: [Proxool-developer] Query execution time

[Bill H. <bi...@lo...>]

Hi Kiran,

On Mon, 2004-01-26 at 20:13, Kiran T wrote:

> Does proxool support, directly or indirectly, determining the execution time 
> for each query?

Yep. Use the ConnectionListenerIF.

> Does onExecute() event triggered at the beginning or the end of execution?

At the end. Although I seem to remember someone else asking for another
event at the beginning of execution too.

onExecute() conveniently passes you the elapsed time for each execution.

It doesn't take account of the time taken to fetch any information from
a Connection, however. For instance, iterating through a large ResultSet
or reading a large Clob.

- Bill

[Proxool-developer] Query execution time

[Kiran T <mah...@ho...>]

Hello all,
Does proxool support, directly or indirectly, determining the execution time 
for each query?

For any given connection, I can get the total time by tracking the time 
taken between getting the connection out of a pool and dumping back to the 
pool but a single connection can have  multiple queries associated with it.

Does onExecute() event triggered at the beginning or the end of execution?

Thanks in advance
Kiran

_________________________________________________________________
Check out the new MSN 9 Dial-up — fast & reliable Internet access with prime 
features! http://join.msn.com/?pgmarket=en-us&page=dialup/home&ST=1

Re: [Proxool-developer] Property maximum-active-time

[Bill H. <bi...@lo...>]

Hi Per Thomas,

On Tue, 2004-01-13 at 16:44, Per Thomas Jahr wrote:

> I'm using Proxool with Hibernate 2.1.1 and in the hibernate.cfg.xml
> there is the Proxool property set to maximum-active-time 500. Is that
> milliseconds? It seems like my application closes databases connections
> to early. Could maximum-active-time = 500 be the problem?

Yep, it's in milliseconds. That important bit of information seems to be
missing from the doc. I'll fix that.

maximum-active-time = 500 means that if the house keeper spots that a
connection is used for more than half a second it will kill it. Since
the house keeper only runs every 30 seconds (by default) it effectively
means the maximum-active-time is anywhere between 0.5 seconds and a tad
over 30.5 seconds.

- Bill

[Proxool-developer] Property maximum-active-time

[Per T. J. <Per...@nr...>]

Hi,

I'm using Proxool with Hibernate 2.1.1 and in the hibernate.cfg.xml
there is the Proxool property set to maximum-active-time 500. Is that
milliseconds? It seems like my application closes databases connections
to early. Could maximum-active-time = 500 be the problem?


Thanks
Per Thomas

Stack trace:

2004-01-13 15:36:45,406 DEBUG [ThreadPoolWorker:Thread-9]
(per@156.116.2.165) (JDBCExceptionReporter.java:36) - SQL Exception
java.sql.SQLException: No operations allowed after connection closed.

Connection was closed explicitly by the application at the following
location:

** BEGIN NESTED EXCEPTION ** 

java.lang.Throwable

STACKTRACE:

java.lang.Throwable
at com.mysql.jdbc.Connection.close(Connection.java:1107)
at
org.logicalcobwebs.proxool.AbstractProxyConnection.reallyClose(AbstractProxyConnection.java:173)
at
org.logicalcobwebs.proxool.ConnectionPool.removeProxyConnection(ConnectionPool.java:413)
at org.logicalcobwebs.proxool.HouseKeeper.sweep(HouseKeeper.java:139)
at
org.logicalcobwebs.proxool.HouseKeeperThread.run(HouseKeeperThread.java:39)


** END NESTED EXCEPTION **


at com.mysql.jdbc.Connection.checkClosed(Connection.java:2741)
at com.mysql.jdbc.Connection.prepareStatement(Connection.java:1373)
at com.mysql.jdbc.Connection.prepareStatement(Connection.java:1353)
at sun.reflect.GeneratedMethodAccessor31.invoke(Unknown Source)
at
sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:25)
at java.lang.reflect.Method.invoke(Method.java:324)
at
org.logicalcobwebs.proxool.ProxyConnection.invoke(ProxyConnection.java:68)
at
org.logicalcobwebs.cglib.proxy.Proxy$ProxyImpl$$EnhancerByCGLIB$$9d354a27.prepareStatement(<generated>)
at
net.sf.hibernate.impl.BatcherImpl.getPreparedStatement(BatcherImpl.java:228)
at
net.sf.hibernate.impl.BatcherImpl.prepareQueryStatement(BatcherImpl.java:61)
at net.sf.hibernate.loader.Loader.prepareQueryStatement(Loader.java:703)
at net.sf.hibernate.loader.Loader.doQuery(Loader.java:184)
at
net.sf.hibernate.loader.Loader.doQueryAndInitializeNonLazyCollections(Loader.java:132)
at net.sf.hibernate.loader.Loader.loadCollection(Loader.java:909)
at net.sf.hibernate.loader.Loader.loadCollection(Loader.java:884)
at
net.sf.hibernate.loader.CollectionLoader.initialize(CollectionLoader.java:69)
at
net.sf.hibernate.collection.AbstractCollectionPersister.initialize(AbstractCollectionPersister.java:284)
at
net.sf.hibernate.impl.SessionImpl.initializeCollection(SessionImpl.java:3133)
at
net.sf.hibernate.collection.PersistentCollection.initialize(PersistentCollection.java:203)
at
net.sf.hibernate.collection.PersistentCollection.read(PersistentCollection.java:69)
at net.sf.hibernate.collection.Bag.size(Bag.java:231)



-- 
Per Thomas Jahr

Re: [Proxool-developer] Why embeed external packages?

[Bill H. <bi...@lo...>]

Hi Claudio,

On Sat, 2004-01-10 at 11:02, Claudio Destro wrote:

> On the countrary, proxool-0.8.3 logs messages using its Jdk14Logger; so I get 
> my messages in a text file while proxool messages go to stdout.
> 
> Moreover, setting the org.logicalcobwebs.logging.Log System property
> (with -Dorg.logicalcobwebs.logging.Log=com.tntstudio.log.ConsoleLogger)
> produces the following stack trace:
> 
> Exception in thread "main" java.lang.ExceptionInInitializerError
>         at com.tntstudio.skyware.Skyware.startup(Unknown Source)
>         at com.tntstudio.skyware.Skyware.main(Unknown Source)
> Caused by: org.logicalcobwebs.logging.LogConfigurationException: 
> org.logicalcobwebs.logging.LogConfigurationException: 
> org.logicalcobwebs.logging.LogConfigurationException: Class 
> com.tntstudio.log.ConsoleLogger does not implement Log

We can easily fix the first issue: getting Proxool to use the standard
property value, but that second one is a bit trickier. Of course, you
could make your ConsoleLogger also implement
org.logicalcobwebs.logging.Log but that is a bit clumsy.

I'd like to have this all working transparently, I really would. It is
starting to look like we should switch back to including JCL as a
separate JAR. 

For the record, I have no compulsion to do the same for the concurrent
and collection classes we have. They have no public interaction at all.
Ditto with cglib.

> I think proxool's 
> internal JCL prevents usage of any other org.apache.commons.logging.Log 
> implementation.

Agreed.

Bill

Re: [Proxool-developer] Why embeed external packages?

[Claudio D. <o1...@ti...>]

Hi Bill,

I use the following commons-logging.properties file to force the Jakarta 
Commons Logging system to use a specific Log:

--BOF--
#
# Instruct the JCL default LogFactory to instantiate our specific
# implementation.
#
org.apache.commons.logging.Log=com.tntstudio.log.ConsoleLogger

# Our logger can be configured by setting few properties in the
# "console-logger.properties" file.

--EOF--

(I've written the ConsoleLogger class just for fun...)

Everything is logged by our class correctly to a text file, also proxool-0.6 
messages.

On the countrary, proxool-0.8.3 logs messages using its Jdk14Logger; so I get 
my messages in a text file while proxool messages go to stdout.

Moreover, setting the org.logicalcobwebs.logging.Log System property
(with -Dorg.logicalcobwebs.logging.Log=com.tntstudio.log.ConsoleLogger)
produces the following stack trace:

Exception in thread "main" java.lang.ExceptionInInitializerError
        at com.tntstudio.skyware.Skyware.startup(Unknown Source)
        at com.tntstudio.skyware.Skyware.main(Unknown Source)
Caused by: org.logicalcobwebs.logging.LogConfigurationException: 
org.logicalcobwebs.logging.LogConfigurationException: 
org.logicalcobwebs.logging.LogConfigurationException: Class 
com.tntstudio.log.ConsoleLogger does not implement Log
        at 
org.logicalcobwebs.logging.impl.LogFactoryImpl.newInstance(LogFactoryImpl.java:526)
        at 
org.logicalcobwebs.logging.impl.LogFactoryImpl.getInstance(LogFactoryImpl.java:279)
        at 
org.logicalcobwebs.logging.impl.LogFactoryImpl.getInstance(LogFactoryImpl.java:250)
        at org.logicalcobwebs.logging.LogFactory.getLog(LogFactory.java:380)
        at 
org.logicalcobwebs.proxool.ProxoolDriver.<clinit>(ProxoolDriver.java:28)
        ... 2 more
Caused by: org.logicalcobwebs.logging.LogConfigurationException: 
org.logicalcobwebs.logging.LogConfigurationException: Class 
com.tntstudio.log.ConsoleLogger does not implement Log
        at 
org.logicalcobwebs.logging.impl.LogFactoryImpl.getLogConstructor(LogFactoryImpl.java:410)
        at 
org.logicalcobwebs.logging.impl.LogFactoryImpl.newInstance(LogFactoryImpl.java:519)
        ... 6 more
Caused by: org.logicalcobwebs.logging.LogConfigurationException: Class 
com.tntstudio.log.ConsoleLogger does not implement Log
        at 
org.logicalcobwebs.logging.impl.LogFactoryImpl.getLogConstructor(LogFactoryImpl.java:406)
        ... 7 more

I know it's silly to write a new SimpleLog as I did :(, but I think proxool's 
internal JCL prevents usage of any other org.apache.commons.logging.Log 
implementation.

Claudio

Re: [Proxool-developer] Why embeed external packages?

[Bill H. <bi...@lo...>]

HI Claudio,

On Wed, 2004-01-07 at 20:47, Claudio Destro wrote:

> I think the problem is in the re-writing of "org.apache.commons.logging.Log" 
> property in "org.logicalcobwebs.logging.Log" (internal JCL does not act the 
> same way as external one).

I understand that. We made the decision to repackage all *internal*
dependencies, but logging is not entirely internal so maybe we need to
make a special case for it. It does handle configuration surprisingly
well and will co-exist with the commons' logging library. But there may
be some specific cases here that we need to deal with.

> So, I understand version problems, but why not simply retain external package 
> original names instead of embeed them into org.logicalcobwebs super-package? 

If we used the same package names then it would not be clear which class
was being used - it would depend on the classpath order. I don't think
that is very satisfactory. Class loader issues are best avoided.

Issues:

1) It's nice that Proxool doesn't have any dependencies at the moment.
Some projects can't avoid them, but as long as we can I'd like to.

2) We don't want different behaviour depending on whether other
libraries get loaded first or not.

3) On the other hand, logging should be transparent. If you setup
Commons' logging in your project then Proxool should react correctly.

Maybe the answer is to fix the precise logging problem you are having. I
have never fully understood that "org.apache.commons.logging.Log"
property. Perhaps it's time I did.

Bill

[Proxool-developer] Why embeed external packages?

[Claudio D. <o1...@ti...>]

These days I was thinkering with an annoying _bug_: while my classes logs - 
using external JCL/SimpleLog - to a text file, proxool-0.8.3 logs - using its 
internal JCL/Jdk14Logger - to System.out.

I think the problem is in the re-writing of "org.apache.commons.logging.Log" 
property in "org.logicalcobwebs.logging.Log" (internal JCL does not act the 
same way as external one).

So, I understand version problems, but why not simply retain external package 
original names instead of embeed them into org.logicalcobwebs super-package? 
I understand this for one or few more classes (see util.FastArrayList, I did 
the same with a *modified* version of catalina.util.FastDateFormat).

This gives users the ability to un-embeed external packages (avoiding useless 
duplication of code), and saves you from version problems (even if I think 
this is not a real problem expecially for packages like Concurrent and 
Logging).

In any case you can distribute all sources needed by Proxool (see Apache 
XmlRpc which boundles uk.co.wilson.xml.MinML and uk.org.xml.* in its source 
distribution). I mean:

proxool-x.x.x-src.jar
    |
    +-- org/logicalcobwebs/proxool/*.java
    |
    +-- org/apache/commons/logging/*.java
    |
    +-- EDU/oswego/cl/dl/util/concurrent/*.java
    |
    +-- net/sf/cglib/*.java
    |
    +-- ...

...Just an opinion.

Claudio Destro <cd...@tn...>

Re: [Proxool-developer] Connection Events

[Bill H. <bi...@lo...>]

Hi Kiran,

On Mon, 2003-12-22 at 17:22, Kiran T wrote:

> I have been trying to make use of ConnectionListener events and I somehow 
> feel that the events are not comprehensive. Just for an example, I was 
> trying to make use of onDeath() to send out emails whenever connection gets 
> "killed" but I didn't want to send emails when these connections were 
> terminated naturally( ie after 4 hours of being in the pool).

I suppose we could send a code that described why the connection was
being destroyed. Off the top of my head, a connection can be killed
because:

a) it has reached its maximum lifetime (default 4 hours)
b) it has failed its house keeping test
c) it has thrown a fatal SQL exception

There might be some other cases too.

> Also, I figured out that the java.sql.Connection parameter passed to the 
> onDeath() is not the Proxool Connection but the native java.sql Connection. 
> Was there any reason to not pass proxool's connection? I normally put the 
> connection in the map while getting the connection and pop out during 
> onDeath() call. Because it was not a proxool connection I had to use 
> ProxoolFacade.getDelegateConnection() and then use it.

Hmm, I can't think of any reason not to set the Proxool connection. I
think the only reason you get a connection (in that method) is so that
you can perform a resource clean up. And I don't see why you can't do
that with a Proxool Connection. I'll have a think about the implications
of this.

> I remember an earlier discussion on splitting onExecute() to 
> onExecuteStart() and onExecuteEnd(). Thats a great idea. Would it be 
> possible to integrate in 0.9.0?

That means a change to the API and is not something we do lightly.
However, we are pre-1.0 and that is something that we are prepared to
change if there is a good reason. Your suggestions sound good to me.
Anybody going to argue against them?

Thanks for your constructive comments, Kiran.

Cheers,
Bill