RP2040 ram issue is back to haunt me

[kjm1102]

A few years ago, when I started getting memory allocation errors, I switched to using .mpy files in lieu of .py, the pico was happy to do a 20k OTA download of a new .mpy Vs 60k for a .py and it was easy enough to use boot.py to import code.mpy to run my programs.

However the memory allocation errors left me wary since I didn't understand how a 60k .py file could not download a new version of itself when the chip reports 188k of ram available at powerup. So every so often I check the userland code updater still works.

I do the check in the IDE with boot.py disabled so I can can run the .mpy file from the prompt by typing import code. The .mpy files are still only ~20k in size & the new version downloads from the website.

Today a .mpy file self started from powerup. It did this because I forgot to disable boot.py which promptly imported the .mpy file. To my great distress the OTA update failed with a memory allocation error. I have no clue why this should be? Why should a program triggered by an import code cmd in boot.py not work while the same program started by importing it manually from the IDE prompt works OK?

[kjm1102]

Spent the last week pruning the .mpy, this is no way to live!
When it got down to ~18.5 k the memory allocation error went away. I still can't get my head around why a microprocessor claiming to have 155588 free ram immediately before the download can't take files bigger than ~18.5k. What am I missing here?