Re: Classloader question

2007-07-18 Thread Christopher Schultz
Juan, Juan Ignacio Garzón wrote: > Is there a way for making a .jar available to all the > applications, but using each application's class loader, so that > static fields are not shared between applications? No. You will have to put the JAR file into each webapp's WEB-INF/lib directory. You cou

RE: Classloader Question

2006-10-11 Thread Fran Varin
Larry, We tried the "shared.loader" suggestion and it seems to be working fine. Thanks very much for the suggestion. Fran Larry Isaacs wrote: > > Since you are using Tomcat 5, check out the "shared.loader" > property specified in the catalina.properties file of your > Tomcat instances. Yo

Re: Classloader Question

2006-10-10 Thread Fran Varin
Thanks for the suggestions! I'll look into these and post back either way. Fran Fran Varin wrote: > > We are running multiple Tomcat 5.5 instances as Windows services. We have > some .jar files that are common between the multiple Tomcat instances. We > have been searching for a way to conf

RE: Classloader Question

2006-10-10 Thread Larry Isaacs
Since you are using Tomcat 5, check out the "shared.loader" property specified in the catalina.properties file of your Tomcat instances. You could change it to use "catalina.home" instead of "catalina.base", or add an additional "path". Cheers, Larry > -Original Message- > From: Fran Var

RE: Classloader Question

2006-10-10 Thread Huy Vo \(hvo\)
- From: King, Patrick [mailto:[EMAIL PROTECTED] Sent: Tuesday, October 10, 2006 10:32 AM To: Tomcat Users List Subject: RE: Classloader Question A possible solution would be to use the analog of a "unix file link" for windows based operating systems. One tomcat distribution woul

RE: Classloader Question

2006-10-10 Thread King, Patrick
A possible solution would be to use the analog of a "unix file link" for windows based operating systems. One tomcat distribution would have the actual jar file, and the other tomcat distribution would have a "file link" to the same file(the file link is simply a "pointer" to the actual jar fil

Re: Classloader question

2006-03-24 Thread Boris Unckel
Hello, Fran Varin wrote: The beauty of our WAS solution is that we can hot deploy various pieces like the jars without having to do anything with the WARs and since we do not have the jars contained in each WAR it makes maintenance much simpler. Depending on the application, this approach makes a

Re: Classloader question

2006-03-24 Thread Fran Varin
The beauty of our WAS solution is that we can hot deploy various pieces like the jars without having to do anything with the WARs and since we do not have the jars contained in each WAR it makes maintenance much simpler. Depending on the application, this approach makes a lot of sense. We do use A

Re: Classloader question

2006-03-24 Thread Reinhard Moosauer
Hi, this sounds like "repackage" would be a huge job. If yes, I would suggest: learn ant If not: You have to reload the apps anyway. So what do you achieve? just my 2 cents. R. Am Freitag, 24. März 2006 16:19 schrieb Fran Varin: > Yes, that is the exact situation we are attempting to avoid in T

Re: Classloader question

2006-03-24 Thread Fran Varin
I tend to agree with your postion. I'm not sure that Tomcat will see the native Windows shortcut. I'll see if we can figure out how to do it. If you have any ideas I'm all ears. It sure would be nice if Tomcat had the ability to have the shared lib see down the hierarchy or be able to put jars so

Re: Classloader question

2006-03-24 Thread Mike Sabroff
I run cygwin (a linux like application/environment) on my windows box at home. When I create a symbolic link in that system and look at it frm the windows side of it, it IS a shortcut. try itI think it may work, but I am not sure that tomcat will see it as cygwin does. Curious. Fran

Re: Classloader question

2006-03-24 Thread Fran Varin
...as in Windows shortcut...I'll have to look into that possibility. -- View this message in context: http://www.nabble.com/Classloader-question-t1332679.html#a3573968 Sent from the Tomcat - User forum at Nabble.com. - To unsu

Re: Classloader question

2006-03-24 Thread Mike Sabroff
SHORTCUT! Fran Varin wrote: aah...now I understand the reason it sounded foreign to me. We are a Windows shop so, I'm not sure we have the same capability. -- View this message in context: http://www.nabble.com/Classloader-question-t1332679.html#a3573581 Sent from the Tomcat - User forum at N

Re: Classloader question

2006-03-24 Thread Fran Varin
aah...now I understand the reason it sounded foreign to me. We are a Windows shop so, I'm not sure we have the same capability. -- View this message in context: http://www.nabble.com/Classloader-question-t1332679.html#a3573581 Sent from the Tomcat - User forum at Nabble.com. --

Re: Classloader question

2006-03-24 Thread Boris Unckel
Hi, > Von: Fran Varin <[EMAIL PROTECTED]> > This approach sounds promising...would you mind elaborating just a little > on > what you're thinking? I'm not sure I follow when you mention using a > symbolic link into WEB-INF/lib. it would require UNIX or LINUX system. A simple symbolic link: ln -s /

Re: Classloader question

2006-03-24 Thread Fran Varin
This approach sounds promising...would you mind elaborating just a little on what you're thinking? I'm not sure I follow when you mention using a symbolic link into WEB-INF/lib. -- View this message in context: http://www.nabble.com/Classloader-question-t1332679.html#a3573086 Sent from the Tomca

RE: Classloader question

2006-03-24 Thread Fran Varin
Yes, that is the exact situation we are attempting to avoid in Tomcat. In our WSA implementation we can simply replace the effected JARS in one location and it is implemented against all WARs running in that server's context. In the illustration you provided below we would need to repackage, or at

Re: Classloader question

2006-03-24 Thread David Smith
Boris Unckel wrote: Hello, Von: Fran Varin <[EMAIL PROTECTED]> Yes, quite correct on your statement regarding using "Application" as the definition. The essence of what we are looking for is a similar behavior with Tomcat. Our over arching goal is to minimize or eliminate the need to have j

RE: Classloader question

2006-03-24 Thread Caldarale, Charles R
> From: Boris Unckel [mailto:[EMAIL PROTECTED] > Subject: Re: Classloader question > > To the mailing-list: If you have an library which has not > the explicit recommendation to put it in common/shared lib > path (i.E. a special JDBC driver where the vendor recommends &g

Re: Classloader question

2006-03-24 Thread Boris Unckel
Hello Dave, > Von: David Kerber <[EMAIL PROTECTED]> > I understand the arguments on both sides, but tend to prefer the ease of > maintenance of what you call "the single point of change in > shared/lib". Is it possible to make this configurable, so both sides > can be happy? Or is that too com

Re: Classloader question

2006-03-24 Thread David Kerber
Boris Unckel wrote: ... To the mailing-list: If you have an library which has not the explicit recommendation to put it in common/shared lib path (i.E. a special JDBC driver where the vendor recommends one to put it into shared) what do you prefer - the single point of change in shared or the i

Re: Classloader question

2006-03-24 Thread Boris Unckel
Hello, > Von: Fran Varin <[EMAIL PROTECTED]> > Yes, quite correct on your statement regarding using "Application" as the > definition. The essence of what we are looking for is a similar behavior > with Tomcat. Our over arching goal is to minimize or eliminate the need > to have jars that are to b

Re: Classloader question

2006-03-24 Thread Fran Varin
Yes, quite correct on your statement regarding using "Application" as the definition. The essence of what we are looking for is a similar behavior with Tomcat. Our over arching goal is to minimize or eliminate the need to have jars that are to be shared by more than one applicaiton (WAR) be copied

Re: Classloader question

2006-03-23 Thread Boris Unckel
Hello Francis, VARIN, FRANCIS A. wrote: As mentioned above, we have used this class for several years in WAS. In that case, the jar that contains the factory resides at the EAR level and can instantiate classes that live in the associated WAR files just fine. The problem only exists in Tomca