This uses a similar sort of registry to this question I asked. Inside a class, I have a registry (in the form of a list this time) of all instances of that class. An instance is initialised with some coordinates, but these will be changed as the user uses the program. I have made a function that resets these coordinates (simplified version)
class Foobar:
_registry = []
def __init__(self, position=[0,0,0], direction=[0,0,1]):
self._pos = position # can change later
self._pos0 = position # stored static copy of initial position
self._dir = direction
self._dir0 = direction # stored initial direction
self._registry.append(self) # adds instance to registry
...
def remove(self):
Foobar._registry.remove(self) # removes instance from registry
def reset(self):
self.remove()
Foobar.__init__(self, self._pos0, self._dir0)
# re-initialises instance with starting parameters
...
def reset_all():
# note: outside of Foobar
for item in Foobar._registry:
item.reset()
Now, this works if I do
>>> A = Foobar([0,0,0], [0,0,1])
>>> A.do_something_that_adds_more_pos_&_dir_vectors()
>>> B = Foobar(... # same sort of thing. C, D etc
>>> A.reset()
>>> B.reset()
>>> C.reset()
>>> D.reset()
Those all reset fine. If I enter into the IPython console
>>> for item in Foobar._registry:
... item.reset()
All the objects reset fine. But if I do reset_all() instead, I find that some of the instances reset, whilst others don't. If I keep doing reset_all() eventually they will all have reset, but not in one go. What's going on!?
Extra info: I append position and direction vectors to these objects. I then plot the paths dictated by the vectors. If I try reset_all() and then try to plot them again, some of them get reset and no longer show up on the plot (as expected), but about half still show up.
