Firmware

[patricklatimer]

• Proper way to use events?
• Do we need any other control registers?
• WhoAmI?
• Best place for device.yml?

[Poofjunior]

For bare-bones Harp communication, this looks good. Delegating the bulk of the work to core1 is clever in that you don't need to rewrite the SPI transactions to non-blocking (i.e: cooperative multitasking) to keep up Harp communication.

I added a few in-line comments.

Additionally, let's plan to add:

• an aggregate register of the 3 sensor values represented as a single register.
  • We need to use an array for now (promote pressure from U32 to Float) since Harp spec doesn't support registers pointing to structs
• a periodic dispatch event, when new data from the sensor causes the device to dispatch a Harp EVENT message from the aggregate register.
• a register to enable periodic sensor dispatch

[Poofjunior]

Delegating the bulk of the work to core1 is clever in that you don't need to rewrite the SPI transactions to non-blocking (i.e: cooperative multitasking) to keep up Harp communication.

Actually, on second thought we might need to consider getting this to run on just one core for heat-output reasons. Let's do a temperature check with the current implementation. If the heat change is noticeably higher, we might need to do a rewrite to just use one core.

[patricklatimer]

Ooh, still need WhoAmI, shouldn't forget to add that.

[bruno-f-cruz]

Ooh, still need WhoAmI, shouldn't forget to add that.

You can go ahead and use 1405 (I have registered it https://github.com/harp-tech/protocol/blob/41941a6f2dc4f49397946544d70aa5d9759a0ed8/whoami.yml#L173). Once you have a firmware you are happy with let me know and I can generate the bonsai interface and device.yml. Thanks!

[patricklatimer]

Okay, so I was going to finish some pyharp upgrades and python tests for this, but it sounds like I'm being pulled elsewhere and it isn't actually blocking the bonsai work so I'll call this ready for review and integration into bonsai if y'all approve.

[Poofjunior]

Ok overall, this looks great! Let's handle these in-line changes and go from there.

Also: that's a rad way to quickly spin up a Harp class for this device on the Python side. It would be super awesome to get it into pyharp eventually.

[bruno-f-cruz]

Ok overall, this looks great! Let's handle these in-line changes and go from there.

Also: that's a rad way to quickly spin up a Harp class for this device on the Python side. It would be super awesome to get it into pyharp eventually.

In case it's useful we are maintaining a parser for yml files here already https://github.com/harp-tech/harp-python/blob/main/harp/schema.py

[patricklatimer]

Alright, I believe everything should be addressed. The python driver probably won't work without the new pyharp driver that handles events (AllenNeuralDynamics/pyharp#4) though.

[Poofjunior]

Woot-woot! Thanks for making all the tweaks!