raise warning when no external drives added

[cjayb]

@cjayb could we perhaps raise a warning when running a simulation with no input drives? And telling users that they must add the drives manually or use add_drives_from_params=True. If they use add_drives_from_params=True, a big fat deprecation warning should be raised.

+1 on both counts.

perhaps we could not add a drive to net.external_drives if in the param file t0 of drive > tstop of simulation?

and

The items in net.external_drives should map 1-to-1 with whatever is specified by a param file or the user via calls to net.add_... so that we don't end up silently loosing track of network drives, gids, etc.

I propose raising a ValueError for drives (from params or otherwise) where t0 > T. The solution is either to fix the param-file (setting weights to zero effectively "removes" the drive), or, preferably, use the new API.

[jasmainak]

I propose raising a ValueError for drives (from params or otherwise) where t0 > T

I think we agreed in #397 that we'll truncate rather than raise an error?

The warning is easy to tackle?

I'm trying to look at low hanging fruits from 0.2 milestone and get them done to prepare for a release

[jasmainak]

probably easily bunched with #410 ?

[cjayb]

Right, now that tstop is a parameter of the simulation, the t0 of the drives should be allowed to be arbitrary. Perhaps a RuntimeError if simulate_dipole is called on a net without drives?

[jasmainak]

yeah that sounds good. Not sure if we want an error or warning ... I remember @ntolley saying there were situations where you'd want to simulate a network without driving it

[ntolley]

Close issue? I think we've moved towards the opinion that less warnings is good, but maybe this one is worth it.

[jasmainak]

I would say it's worth it and an easy issue to tackle for a new contributor.