Iter_connection on edges for subset of nodes crashes

[arnaudon]

If I call edges.iter_connections(ids) where ids is in a single population of a multi-population circuit, this for loop

snap/bluepysnap/edges/edges.py

Line 246 in 835165d

yield from pop.iter_connections(

crashes here:

https://github.com/BlueBrain/snap/blob/master/bluepysnap/nodes/node_population.py#L342

as it tries to get the connections in another population first.

I just wondered if it would be possible to have a smarter iter_connection to not crash, but return all connections, regardless to which populations the source/target cells belong to.

[eleftherioszisis]

Given that you pass ids e.g. [0, 1, 2, 3, 4, 5] to the function and there are multiple edge populations with node populations starting from 0 and with different sizes, I am not sure what the expected result should be.

Should it return all the connections from every edge pop that has node ids within that range, skipping out-of-size ids if any?

That doesn't sound practical for any analysis whatsoever. On the contrary, a behavior like this could be quite dangerous as it will return values where intuition would expect an error.

I would use the population level iter_connections at all times. I am not sure what is the purpose of iter_connections at the edge storage level when we have multiple edge populations with node populations of different sizes.