Trying to get OpenCMS working with OC4J 10.1.2 (or 10.1.3 dp4) seems to be quite challenging task. What makes this very interesting is that I don't seem to get any errors or exception messages in any of the logs despite of enabling full debugging in log4j.properties file.
There were definitely something wrong with the installation of OpenCMS on OC4J. The installation failed on NPE but system somehow managed to start up. Perhaps some of the modules got installed correctly. I can see that some of the config files point to Tomcat setup (like pointing to ROOT -default web application), so there might be something Tomcat specific behind the scenes that affect the application execution on other containers.
Anyways, to fork the problem area I installed OpenCMS using Tomcat (which works fine) so that I got the repository installed correctly on my Oracle 10.1.3 database. What I did then was I copied all the OpenCMS config etc. files under Tomcat webapps to the OC4J directory to make sure everything is as it should be after the installation. What happens is that systems starts up nicely, the log file compared to Tomcat looks exactly the same (used UltraEdit file diff'er to find out the differences). But what happens next after entering login (Admin/admin), is not something I am looking for. The login screen keeps popping up all the times and the workspace isn't shown. No error messages, no exceptions, no beeps, no nothing.
One step closer to despairness was when I downloaded the whole OpenCMS sources tree on my JDeveloper and re-built the whole application from sources. Now the only thing to find out was to remotely debug the OpenCMS server using JDeveloper's remote debugger and starting OC4J in debugging mode. Ok. This should work... or should it? I set the debug break point and debug the application line-by-line. For some reason the debugger freaks out in the critical moments that should point out the problem area. What is going on?
I also see that other people face the same problem when looking at OpenCMS mailing list.
Battle continues. Having the sources and detail debugging should eventually reveal the source of the problem. Let's see how long this takes.