sPyNNaker no clear mapping between python and c code recording regions

It ust to be that a variable had a given recording id and that it was handed down from python to the c code, thus avoiding mismatch between python and c. now the c has a hardcoded set of #defines which flag the index's, but there is no such fixment in the python, it only occurs due to a ordering of a list. this should instead be sent down to the c code, so that future changes do not cause this to crash.

moved from https://github.com/SpiNNakerManchester/SpiNNFrontEndCommon/issues/365#issuecomment-433142275

Oct 26 '18 08:10 alan-stokes

If it "ust to be" that was MANY years ago!

Oct 26 '18 08:10 Christian-B

Unclear what you want changing.

Apr 09 '19 15:04 Christian-B

The hard-coded set of defines in C exactly matches the order of the hard-coded list in Python... I’m not sure that adding an extra layer to this on the off-chance that there might be some variable added to the list at some point (what would the new variable be, anyway...?) would be a good idea. It’s no different really from a neuron model where you have to send the neuron parameters onto the machine in the same order as they have been defined in the C code...

Jul 29 '19 16:07 andrewgait

think this has been subercedded by neuron recording. as matrix vs bitfield regions are no longer special

Sep 27 '19 10:09 alan-stokes

Assigning to @andrewgait who I expect will close based on his comment from 29 Jul 2019

Mar 22 '21 16:03 Christian-B

We gained the extra layer as part of allowing references to regions (which enables sharing synaptic matrices between cores on the same chip). Is that "clear"? 🤷‍♂️

Sep 27 '23 12:09 dkfellows