[hibernate-issues] [Hibernate-JIRA] Created: (HHH-5066) EJB JARs are always re-downloaded on a J2EE application using Java Web Start

[Toni Bennasar Obrador (JIRA)]

EJB JARs are always re-downloaded on a J2EE application using Java Web Start
----------------------------------------------------------------------------

                 Key: HHH-5066
                 URL: http://opensource.atlassian.com/projects/hibernate/browse/HHH-5066
             Project: Hibernate Core
          Issue Type: Bug
    Affects Versions: 3.3.1
         Environment: Ubuntu 9.10 / Hibernate Annotations 3.4.0.GA / Hibernate 3.3.1.GA / Hibernate Commons Annotations 3.1.0.GA / Hibernate EntityManager 3.4.0.GA / mysql  Ver 14.14 Distrib 5.1.37, for debian-linux-gnu (x86_64) / Java(TM) SE Runtime Environment (build 1.6.0_17-b04)
            Reporter: Toni Bennasar Obrador
         Attachments: InputStreamZippedJarVisitor.java

Hello folks.

I have an J2EE application that uses JPA through hibernate to access to the database, directly, without a J2EE server. The application works fine when its running on the development enviroment. We deploy the application with java web start (jws), and here comes the problem. When its started from a remote workstation, the application stands frozen 2 minutes until it starts. Looking at the jws console i see that get frozens here:

...
XXX  INFO Environment:627 - using JDK 1.4 java.sql.Timestamp handling
XXX  INFO Version:14 - Hibernate Commons Annotations 3.1.0.GA
XXX  INFO Version:16 - Hibernate EntityManager 3.4.0.GA

< frozen HERE....

XXX  INFO AnnotationBinder:419 - Binding entity from annotated class: fac.vta.MovimientoEdicion
...

I have used the tcpdump tool, and i see that the application each time its started downloads AGAIN all our EJB jar files from the web server (they were downloaded the first time we started the application with jws). I did a debug and i see that they are downloaded when the function "doProcessElements" of org.hibernate.ejb.packaging.InputStreamZippedJarVisitor is executed, at this line:

jis = new JarInputStream( jarUrl.openStream() );


Here is my explanation: When you use jws, to this function arrives a jarUrl of this kind: "http://foobar.com/hello/file.jar". This URL was filtered by JarVisitorFactory.getJarURLFromURLEntry class. Initially it was "jar:http//foobar.com/hello/file.jar!/". When you get a url.openConnection() of a "jar:http//foobar.com/hello/file.jar!/" URL, its handled by jws and returns a jws internal CachedJarURLConnection to the locally downloaded JAR, but if you do the url.openConnection() from a "http://foobar.com/hello/file.jar", it returns a HttpURLConnection, and the JAR is downloaded again...

The natural solution should be change to 

if( url.startsWith("http:") ) {
   URL urlJarCheck = new URL("jar:" + url + "!/");
   jis = new JarInputStream( urlJarCheck.openConnection() );
}

but this don't work due to jws sucks. It throws a "no entry name specified" IOException. It seem cannot return a InputStream of the whole jar file at any way i have tried.

The only way i have found to make it work is the one I have put on the attachment. 

I have not tested it with the core v3.5.0, but i have checked the code and it seems having the same problem.

If this a good solution you should pay me a beer. lol.


-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira