Currently, to support NumPy's unconventional == operator (returns array of bools, not a bool), the constructor for taxon_info reaches into Python, grabs the object's classes __eq__ operator, and then uses a try-catch statement to swap in the numpy.array_equal operator.
For a given systematics manager, however, the == should be the same for all taxon_info objects. Thus, there is almost certainly a way to store single copy per systematics manager and avoid having to reach into Python every time the constructor is called.
Possible solutions:
The equals operator could live in the wrapped systematics object (need to figure out how to give taxon_info objects access to it).
The taxon_info object could have the equals operator as a static member (tricky, because it can seg-fault on destruction)
Currently, to support NumPy's unconventional
==operator (returns array of bools, not a bool), the constructor for taxon_info reaches into Python, grabs the object's classes__eq__operator, and then uses a try-catch statement to swap in thenumpy.array_equaloperator.For a given systematics manager, however, the
==should be the same for all taxon_info objects. Thus, there is almost certainly a way to store single copy per systematics manager and avoid having to reach into Python every time the constructor is called.Possible solutions: