matrix.wait doesn't block?

[BrianPugh]

it seems like the C implementation of the matrix.wait method immediately returns, resulting in a super tight primary loop. The python matrix.py code does not have this issue.

Is the C implementation public anywhere?

[xiongyihui]

The C implementation of the matrix.wait will return when a new event (keyup or keydown) occurs. The reason may be that C is faster than Python.

The source code is at https://github.com/xiongyihui/circuitpython/tree/m60

[BrianPugh]

thanks, i'll check that out and raise an issue in that repo if my report is still valid.

[BrianPugh]

ok so I cannot open issues in that repo for some reason. Basically I believe the issue is this:

1. When port_sleep_until_interrupt (or port_idle_until_interrupt in newer circuitpython) is called, sd_app_evt_wait blocks in lower power mode until an interrupt happens.
2. We expect that interrupt to be the interrupt from the timer setup before hand to wake it up after 20mS or 1000mS (or whatever depending on the situation) or for the interrupt to be from a button press.
3. However, I think USB and BT events also trigger this to unblock. I don't really have a way to debug this.
4. Essentially, this results in unblocking nearly immediately, rather than after the expected timeout.
5. wrapping this in a while loop and toggling a flag in the interrupts semi-solves this situation, but this causes a bunch of other issues, like with circuitpython operating properly.

Any thoughts? Another solution is to just wrap this wait code in a while loop in python, just might have battery life impacts.

You can see my circuitpython fork of @xiongyihui repo here:
https://github.com/BrianPugh/circuitpython/tree/m60
I merged (and fixed conflicts) upstream adafruit circuitpython/main into it for bugfixes and I was having linker issues that upstream resolved.