As of today, Numba doesn't support optimization of generic Python objects. This limitation, however, doesn't have a huge impact on numerical codes as they usually involve arrays and math operations exclusively.
Nevertheless, certain data structures are much more naturally implemented using objects; therefore, Numba provides support for defining classes that can be used and compiled to fast, native code.
Bear in mind that this is one of the newest (almost experimental) features, and it is extremely useful as it allows us to extend Numba to support fast data structures that are not easily implemented with arrays.
As an example, we will show how to implement a simple linked list using JIT classes. A linked list can be implemented ...
