Andrew Rowley

Hmm OK, I guess this means that it works like a normal Python list in that case, which would be more of a reason to keep the ID object in...

I think it does in master, but I think it may change in the reduce_overhead_data_spec branch...

Interesting - it should be easy enough to implement this (I think), though the consequences on how many connections are going to be made is even more exciting...

See here: https://github.com/SpiNNakerManchester/PACMAN/blob/master/pacman/utilities/algorithm_utilities/routing_algorithm_utilities.py#L543 and here: https://github.com/SpiNNakerManchester/PACMAN/blob/master/pacman/utilities/algorithm_utilities/routing_algorithm_utilities.py#L558

Note, the alternative is to make the code error when the target is not a local_only. At present it results in an RTE in the worst case!

I think this is ready now - the models work as they are

Jenkins is green so merging!

I think the behaviours are correct; the clear means clear everything up to now on the variable I have chosen. This basically says I have done with the data before...

Ideally, even in the 2D case it would start with 256 neurons per core here I suppose. The issue is down to the fact that we are interpretting PyNN constructs...

I was trying to do too many things here I think! I think we have fixed the core issue in another branch, so this is here to remind me about...