It works normally if I revert to CP 3.0.0-alpha.6

It also works normally if I just remove PR #804

looking at main.py I see that now the FS is write protected to python for all builds:
https://github.com/adafruit/circuitpython/blob/master/main.c#L330

Is that intended. When I revert #804 I am able to write to the FS from within python. For example:

import os
os.rename('foo.py','bar.py')

works

for the non atmel-boards (nrf52, esp8266) should the FS be write inhibited?
Is this part of the problem here?

ah - simple workaround
removing

#define CIRCUITPY_BOOT_OUTPUT_FILE "/boot_out.txt"

from mpconfigport.h fixes the problem for the nrf52. It now enters the REPL as expected with the current master.

This issue is still present after #811.
No response from REPL in terminal session or via ampy.

I can still make it work by removing the define for CIRCUITPY_BOOT_OUTPUT_FILE in mpconfigport.h

FYI - I noted that the ESP8266 mpconfigport.h does not define CIRCUITPY_BOOT_OUTPUT_FILE and it continues to function normally. Is it intentional that it not be defined and boot_out.txt not be created for the ESP8266? Should it also be this way for the nrf52?

The nrf52 build should model itself after the SAMD port. The ESP8266 version isn't using the shared supervisor code yet.

Just saw this (no email, not sure why, sorry) ... I'll see if I can reproduce.

@microbuilder #813 provides a workaround - It has been merged in to master so the nrf52 does boot to REPL now -- the long term fix is still in progress.

the issue is caused by the mp_hal_stdout_tx_str() is called before serial_init() is invoked due to CIRCUITPYTHON_BOOT_OUTPUT_FILE. The fix is skip output to serial if it is not inited.

microbuilder@53593d1

@hathach Thank you for the fix. Please make a pull request with the change and we'll merge it in.

I think this is fixed.