[Bug] Commit 660d728 breaks EV3 UART REPL

[ArcaneNibble]

For some reason which I haven't fully investigated yet, commit pybricks/pybricks-micropython@660d728 causes the EV3 REPL on the debug UART to stop working. The system is not locked up as the USB echo test (which is run in a pbio process) still works and the watchdog timer is still being petted.

Testing the commit before this hash will drop to the >>> prompt as expected, but any commit after this won't print any console output at all.

[laurensvalk]

Thanks for reporting. I hadn't gotten around to testing on EV3 yet. I'll have a look and restore it.

Now that EV3 is a bit closer to a functional state, we can eventually also drop the boot-into-REPL hack and use the normal program flow.

[laurensvalk]

Fix pushed. The issue was that auto-starting the REPL still tried to run the UI instead of skipping it. (It doesn't run on EV3, since it was waiting on Bluetooth to become ready.)

The program-selection process is currently still very much tied to the existing hubs. #2255 makes some steps towards generalizing it, but I've made an effort to make it work the same after refactoring for now so we didn't change too much at once. The next steps are work in progress.