Meeting Feb 2¶

Concerns from last week¶

AVR might lose serial communications that come in when interrupts happen
- If we're using interrupts for timing (or wireless, or anything), this could be a problem
- UART might be a solution to this
- We need to check on this, it might all be hocus pocus
- If necessary, we can modify ROS-serial to resend corrupted data (this looks possible)

Abe's computer is borked for the moment, no major updates from him
- He'll fix it and get back to us
Serial protocol seems relatively lightweight, but Abe hasn't done enough research to give it a firm thumbs-up or thumbs-down
The client library is in Python, which is acceptable (dependent on performance), but not preferred
- We may have to rewrite it to C++

We are going to avoid Arduino's software if at all possible
- It doesn't seem to be tremendously useful
- It would also suck to implement
Homework: find examples that aren't Arduino. Also, datasheet.

Pretty straightforward
Homework is to get a definitive reference
- "Remind me later" - azl
- Bribe him with candy

We can set pins on the AVR to trigger state changes and/or set data to the zigbee
- To send stuff, load stuff into the buffer, set the "okay, GO" pin, get an interrupt on success
- Receive stuff by setting "receive" pin and reading
There exists a new wireless library (called 'wireless-new') that got mostly-written for colony3 and then forgotten
- Goals were to improve reliability (ACK function, toggleable), more intuitive function names and levels of indirection, and more functionality (customizable stuffs)
- It got mostly written and stuck in debugging phase
- We can probably reuse most of the code, and most of the spec
Homework:
- Find, update, and post specification and documentation for the old library

Big idea: wireless should be run entirely from the AVR
- ARM should only see a higher-level interface, have a read/write buffer, and be able to set configs