Unblocking analogbufio

[timchinowsky]

Currently analogbufio.BufferedIn blocks until the number of samples requested have been collected. I'm working on a non-blocking ring buffered version which will enable continuous data transfer. It could work like this:

>>> length = 60
>>> mybuffer = array.array("H", [0x0000] * length)
>>> rate = 10
>>> # create a RingIn object and start reading the ADC into it via DMA
>>> adcbuf = analogbufio.RingIn(board.GP26, sample_rate=rate, mybuffer) 
>>> adcbuf.count
25
>>> adcbuf.count  # this will keep going up and up at 10 Hz
37
>>> adcbuf.last  # this is the last index of the buffer that was written to
55 
>>> adcbuf.last  # it will wrap around according to the length of the buffer
10
>>> mybuffer[adcbuf.last]  # this will return the most recent ADC conversion (as of the time that adcbuf.last executes)
124 
>>> del mybuffer # needs to tolerate this, so we make sure the RingIn still refers to the buffer object
>>> adcbuf.ring[adcbuf.last] # maybe like this
>>> 127

Or the RingIn could allocate the array itself, e.g. adcbuf = analogbufio.RingIn(board.GP26, sample_rate=rate, length=length) Wondering what advice you could offer regarding creation and management of persistent objects like this?

[tannewt]

Generally I like the model of allocating the large chunk of memory outside of the object. That way you can reuse it in a different object if you really want to.

In the Python API, I would expose the indices though. Instead, make an iterator available to get the next value. Take a look at PulseIn for an example.

I'd also suggest adding this to a new module so we don't need to include it everywhere analogbufio is.

[timchinowsky]

Closing in favor of #2676.