Repository navigation
Accepting classes with __call__ methods implemented, not only functions #342
Description
Hi,
I was writing some architecture to leverage PyGad and I came across a problem.
Basically I am using a class architecture and would like to have something like this:
class OnFitness(BaseGeneticClass):
def __call__ (ga_instance: pygad.GA, population_fitness):
# do stuff
on_fitness = OnFitness()
ga_instance = pygad.GA(..., on_fitness=on_fitness)
ga_instance.run()But I can't do that, because in the current implementation, to check the input parameters we have a check on the 'code' property, not for the call one. In this way, we can only use functions on gad, while this simple change to accept classes would be very beneficial to the library.
Thanks!
Reacted by Ahmed Gad
Activity
ahmedfgad commentedon Mar 28, 2026
Thanks @arturossouza for the suggestion. This is going to be supported in the next release.
PyGAD 3.6.0
ahmedfgad commentedon Oct 8, 2026
Callable class instances are now accepted for the fitness function and callbacks. Argument validation uses their callable signatures instead of requiring
__code__. I verified a run using callable instances for bothfitness_funcandon_fitness; both were invoked successfully. See current implementation. I am closing this as implemented.