In Native mode, also known as !DotNet mode, we translate Java syntax and constructs as well as calls to Java libraries to C#/.Net.

Test cases may fail because replaced methods have different behaviors. So correction is required:
 * Inspection and modification of the resulting code with Adapter classes (see ReaderRead)
 * Stubbing some classes to be implemented from scratch (see ClassifierResourceClass)
 * Using helper classes instead of calling .Net libraries directly (see StringSplit)

Janett can currently translate ClassifierForJ in native mode, resulting code will be built instantly, and with 3 manual changes all test cases pass.

In Janett, we prefer to develop Native mode over EmulationMode and VirtualizationMode.