Agent.die()

[catherinedevlin]

Continued from #1354. See also #2149 for the remainder of that PR's methods.

I'm less confident that this PR should me merged, since

• It only joins together two method calls that already exist anyway
• It creates a confusing failure mode (if a model is successfully removed from the schedule, but removal from the space fails, should it be added back to the schedule?)
• If it's called for an Agent in a Model without a Space, perhaps the second failure (to remove the model from space) should not be considered a failure at all? In that case, though, it should really just be a call to self.model.schedule.remove(self)

[EwoutH]

Thanks for this PR! A convenient, complete and robust method to remove agents is certainly useful.

I'm only wondering if there aren't redundant elements in this PR. @quaquel if I'm correct, the current weakrefs already do a large part of the heavy lifting, right?

Also, we have to make clear how this differs from Agent.remove():

mesa/mesa/agent.py

Lines 67 to 70 in 6507f51

	def remove(self) -> None:
	"""Remove and delete the agent from the model."""
	with contextlib.suppress(KeyError):
	self.model.agents_[type(self)].pop(self)

And maybe check if it needs to be removed from AgentSets the agent is part of:

mesa/mesa/agent.py

Line 316 in 6507f51

def remove(self, agent: Agent):

[EwoutH]

@catherinedevlin would you like to continue with this PR? If so we can discuss the differences between die and remove.

[EwoutH]

Closing for now due to lack of response. If anyone wants to work further on it, feel free to leave a message here!