INCONVENIENCES
^^^^^^^^^^^^^^

1. Because of java serialization issues, all the servers should run the
   same jdk, otherwise sometimes objects are incompatible when transferring
   from one to another. The clients should also run jdk's compatible with the
   ones on the servers. (jdk1 - jdk1, jdk2-jdk2 etc)

2. Generally, objects sent from one entity to another should be either basic
   jdk types (string, hash etc) OR should be accessible on all the involved
   platforms (initiator entity jdk, jdk of jvm servers through which the object
   goes through, destination entity jdk)

   Thus, creating a new, unknown object, and sending it out would normally 
   NOT work. This is not such a big problem as it looks like but should be
   taken into account.

3. The server has no control (the current release) on the size of the objects
   that are going through it ... this might lead to memory exceptions on the
   server's side. Currently java doesn't provide coherent means of overcoming
   this problem.

   We are working on a fix. It should become available soon.