Rsa core redesign

[amichair]

Redesign of the RSA core module's import and export registration implementations as well as the service factory and resource management - to untagle the old convoluted design and make it more simple and intuitive, while fixing multiple incompatibilities with the spec that are now more straightforward to implement.

The details are in the JIRA tickets.

[amichair]

Regarding direct access to the Shared class fields:

• This is a private class, so encapsulation via getters/setters is moot - it's all internal implementation details anyway, so doesn't conflict with OOP design principals. This is similar to a class accessing its own private fields directly and not via accessors, which is how the majority of classes are written in OOP (I made up that statistic, but stand by it :-) )

• This is very common even in the Java platform itself - nearly all of the collections we know and love access their inner class fields directly, along with many utilities: HashMap, TreeMap, LinkedList, Hashtable, regex.Pattern, ThreadPoolExecutor and many more (these are just the first ones that came to mind). If it's good enough for Java it's good enough for us ;-)

• Adding getters+setters to all the fields would add 30-40 sloc to the class without a clear benefit - all else being equal, I would prefer less code to do the same functionality.

Is any of this convincing?

[alien11689]

I know that the class is private, I am not blocking it. Maybe I am too focused on business applications or I use lombok so the number of lines is not growing

[amichair]

I rebased on current master and renamed the methods. Other than the encapsulation thing, are there any other issues? Are the rest of the design/impl changes good?

[alien11689]

Beside the encapsulation rules it's ok so approved

[alien11689]

Feel free to merge but please squash the commits or add jira id to the latest commit message

[amichair]

If you meant the first commit then I added a JIRA for it. Not sure what exactly to say about it in the ticket, it's just routine maintenance refactoring within the class, but also not directly related to the changes in the other commits so doesn't feel right to squash them together.

[alien11689]

Let's merge then