SpikeSourceArray ignores new spike_times

[mschmucker]

When setting the spike_times for a SpikeSourceArray? in neuron, only the first setting is effective. New spike_times set at a later point in the simulation with ID.set_parameters() or Population.set() are sored in the python model, but the corresponding neuron mechanism is not made aware of them. The problem seems to be a setting in the INITIAL block in vecstim.mod

net_send(ping+1e-9, 2)

which has the consequence that the mechanism never checks for the availybility of a new spike_times vector.

Additionally, in the PARAMETER block, I set

ping = 1 (ms)

Find attached a modified vecstim.mod which fixes the problem.

[mschmucker]

Oops, there's no

[code]

idiom in tracwiki. Sorry...

[apdavison]

It looks as though this doesn't work in any of the other back-ends, either. I think the reason I originally set the default value of ping to zero was to avoid the overhead of processing the ping event for the common case when the spike_times are not updated. The user was expected to set ping manually.

I have accepted mschmucker's changes (see r717), but the ticket should remain open until the other back-ends are fixed. It might be a good idea to add a ping_interval parameter to SpikeSourceArray to explicitly place this under user control (since the user knows how often the spike_times will be updated and so can set ping to a large value).

[apdavison]

Just realized that ticket:166 is essentially a duplicate of this one, and since I've fixed that, I can close this.