You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It would be really nice to have a test framework for all the basic functions that we have in this repository. There are already some assignment verification routines which are veyr useful for testing new algorithms (https://github.com/ghamerly/fast-kmeans/blob/master/src/kmeans.h#L46), but they are slow and must be compiled in/out.
These new things should be a set of unit tests that interfere as little with the existing code as possible. Things that live in general_functions.h should all be tested, including distances and initialization routines. Dataset should have tests. Any subclass of Kmeans should support such a unit test for itself. Then we can go through instances of subclasses of Kmeans and test them all using virtual methods.
The text was updated successfully, but these errors were encountered:
It would be really nice to have a test framework for all the basic functions that we have in this repository. There are already some assignment verification routines which are veyr useful for testing new algorithms (https://github.com/ghamerly/fast-kmeans/blob/master/src/kmeans.h#L46), but they are slow and must be compiled in/out.
These new things should be a set of unit tests that interfere as little with the existing code as possible. Things that live in general_functions.h should all be tested, including distances and initialization routines. Dataset should have tests. Any subclass of Kmeans should support such a unit test for itself. Then we can go through instances of subclasses of Kmeans and test them all using virtual methods.
The text was updated successfully, but these errors were encountered: