web USB reset of micro:bit upsets data logging programs

[jaustin]

When data logging, the micro:bit has a timestamp which is 'time since boot' - this is nice for quickly previewing data.

If you plug in a battery and then do things over a long period of time the time stamps are reliable, except if you are using MakeCode and web USB - in which case the micro:bit gets arbitrarily reset in a set of cases - certainly when you plug the micro:bit back in while web USB is connected.

The result is that your data doesn't have continuous time stamps.

Is it possible to avoid resetting the micro:bit so often with web USB?

CC @martinwork who noticed that beta does this more than live in #4676 (comment)

[martinwork]

Related: #3418, #3357

[pelikhan]

This is a daplink issue, it resets when setting the baudrate.

[jaustin]

Do you have to set the buadrate? If MakeCode is always 115200 and the micro:bit defaults to that, is it actually necessary?

[pelikhan]

We don't know if the user flashed a program using another environment or changed the baudrate

[jaustin]

But if they have serial at a different baudrate, MakeCode doesn't support that anyway (eg #3537) , and if they changed the baudrate somewhere else, that's a pretty uncommon use-case to drive resetting everyone's board (and therefore resetting all timestamps on data logging programs).

[pelikhan]

Any change in this space would require testing accross (1) V1 vs V2 (2) data logger vs not data logger (3) all versions of DapLink in circulation.

[pelikhan]

Added flags to skip those resets: 347f1e1

[abchatra]

Fixed

[abchatra]

@jaustin what is the plan of action here?

[abchatra]

We reset while setting the baud rate for logging.

[abchatra]

For the next release get rid of the second reset.

[pelikhan]

Requires a hardware test rig with all versions of DAPLink to validate behavior.

[abchatra]

No actionable thing for makecode here.