Let's see the next situation:

- User has a Java-Platform Model UML project 
- User creates a Java project
- User generates code from the UML project to the java project
- User explores the generated project and modified some classes
- User do not reverse engineering of the java project 
   (maybe he forgets or may be he has another reason)
- User changes some elements in the model
- User generates code from the UML project to the same java project 
  
  For now it is planned to backup the existed java source code,
  clean the java project  and generate the new java code from the UML project.
  Note that in this case all user changes will be cleaned.
  For example user has to copy method's bodies from the backup
  and copy them to the java project. 
  So how many times do you do code generation so such times
  you need to copy bodies from backup to your project.


  It is better to do the following to keep java code contents:
  -  Generate code only for elements which were changed in UML project
  -  It needs to indicate was java source changed or not before 
      code generation.
   For example if user generates code and changes the code,
   the corresponded UML element should change own color.
   In this way user checks do he wants to generate code for the 
   element or he needs to do RE before it.