WILCARD_VALUE not in closure

[jdtsmith]

I'm getting this error occasionally:

Traceback (most recent call last):
  File "/flash/lib/scron/week.py", line 74, in _next_step
  File "/flash/lib/scron/week.py", line 61, in get_next_pointer
  File "/flash/lib/scron/scount.py", line 255, in get_next_pointer
  File "/flash/lib/scron/helpers.py", line 118, in items
AttributeError: 'closure' object has no attribute 'WILDCARD_VALUE'

Thanks for scron.

[fizista]

Strange error.

On what device, you get that bug?

What version of micropython?

Could you add the following code before the 118 line in scron/helpers.py file:

  for attr in dir(self):
    print("obj.%s = %r" % (attr, getattr(self, attr)))

Maybe it will tell us more about what is going on there.

[jdtsmith]

Thanks. I'm using Loboris MP on ESP32. In fact I found that scron does not run well at all under that port on the device, leading, in addition to occasional closure errors, to lots of CPU halts apparently related to memory handling, ala:

Guru Meditation Error: Core  1 panic'ed (LoadProhibited). Exception was unhandled.
...
assertion "VERIFY_PTR(ptr)" failed: file "/home/LoBo2_Razno/ESP32/MicroPython/MicroPython_ESP32_psRAM_LoBo/MicroPython_BUILD/components/micropython/py/gc.c", line 594, function: gc_free
abort() was called at PC 0x40117f23 on core 1
...

etc.

These halts were sporadic — sometimes immediately on import, sometimes later. I couldn't identify any particular step or sequence of actions that produced them. In the end I had to abandon scron and write my own simpler loop-polling based scheduler (borrowing the pointer idea, thanks). This has eliminated all such halts.

As to debugging this, I'd guess the issue will be that the loboris ESP32 MP port is no longer supported, so if it contains memory handling bugs that scron tickles, no one is there to address them.

Sadly, the loboris port has major functionality that I can't find anywhere else: display support for common packaged displays like ST7789, background threaded and efficient FTP client, multiple built-in as well as user-compilable font support, and much more. Given the development style, sadly there isn't much appetite to port any of this to mainline.

[fizista]

Thank you for that information. So I thought this problem was related to some MicroPython port.

Maybe someone will discover the cause of these problems. Until then, I'm closing this bug report.