long-lived UART buffer still not working right

[dhalbert]

Following on #1056 and #1080, it appears that PR #1080 still has issues, depending on whether or not the code is imported from a file:

Tested on commit 2e80f37 on an M0 Adalogger:

REPL:

Adafruit CircuitPython 3.0.0-4-g2e80f37 on 2018-08-08; Adafruit Feather M0 Adalogger with samd21g18
>>> import board,busio
>>> uart = busio.UART(board.TX, board.RX)
>>> uart.write(b'\xbb')
1

But in a file:
file test_uart.py:

import board, busio
uart = busio.UART(board.TX, board.RX)

Now in REPL

Adafruit CircuitPython 3.0.0-4-g2e80f37 on 2018-08-08; Adafruit Feather M0 Adalogger with samd21g18
>>> import test_uart
>>> test_uart.uart.write(b'\xbb')
0

Nothing is written (see the 0 return above). Was originally testing with a Saleae, and confirmed that nothing gets written.

tagging @jepler

[jepler]

I'll try to take a look at this tonight. If I came provide a resolution then maybe the change needs to be reverted for now

[dhalbert]

The build that I tested against was too old, and did not include #1080. So I believe this is a red herring. Tested against cac760a, and UART seems to work fine.

@robomike could you retest with this build?
https://adafruit-circuit-python.s3.amazonaws.com/bin/feather_m0_adalogger/adafruit-circuitpython-feather_m0_adalogger-20180809-cac760a.bin

[jepler]

I don't have a Feather M0 datalogger, so I tested with a Feather M0 Express instead.

In order to make a test of both send and receive without getting out the 'scope, I placed a 220Ω resistor between the RX and TX pins.

With the above-shown content in test_uart.py, I opened the repl:

Adafruit CircuitPython 3.0.0-7-gcac760aa6 on 2018-08-20; Adafruit Feather M0 Express with samd21g18
>>> import test_uart
>>> test_uart.uart.write(b'\xbb')
1
>>> test_uart.uart.read(1)
b'\xbb'
>>> 

While testing this, please actually power cycle (don't just use the reset button) before uploading a fresh firmware. It may be superstition on my part, but while originally debugging this problem I believe I had experiences where the 'bad' state that the buggy UART put the microcontroller into survived the soft-reset caused by the reset button.

[dhalbert]

@jepler Thanks for vetting. I think we can close.