MicroPython Forum (Archive)

I don't understand what you mean (I also don't know a lot about Make). The rules are such that OBJ depends on qstrdefs.generated.h, which depends on QSTR_DEFS_COLLECTED (among other things) which depends on qstr.split which depends on qstr.i.last. So the body of the qstr.i.last rule gets ran first, ...

makeqstrdefs.py is called with arguments including the compiler and flags (i.e. QSTR_GEN_CFLAGS), which makeqstrdefs.py uses to do the preprocessing by running the compiler with those flags (you can see that when running make V=1 btw)

so the question is: how does building the qstrdefs work with a circular dependency like this? It's not circular during preprocessing which generates said header, because it's inclusion is guarded by #ifdef NO_QSTR see mkrules.mk (or equivalent genhdr.targets for msvc): # QSTR generation uses the sa...

So to confirm the process in a CI pipeline for a project like mine would be Yes that's it. It's also exactly what MicroPython does on Github Actions (and previously Travis). However, there's no readily available STM32 emulator I think, so ci.sh doesn't run STM32 tests. In your case you'd build the ...

And it seems that these errors should be surpassed, however they are not. They are: -Wno-error means 'do not make this warning an error', not 'do not show this warning'. Is it possible to add to the MAKE command so that I can suppress all warnings that are not errors ?? Not that I know off but you ...

Not sure what you are asking exactly (depends on what your 'micropython project' is), but maybe this helps: in the MicroPython source code there's tools/ci.sh which does everything wrt building and testing. It's used by the github CI (see https://github.com/micropython/micropython/actions for output...

def read(addr) should be def read(self, addr)

Sorry I don't know the ESP-IDF architecture so cannot pinpoint the exact reason (various possibilities include this is another thread, or this is an interrupt, or this runs even before MicroPython had been initialized, etc). The 'standard' workaround if allocation is not possible, is preallocating (...

It's not implemented in the C code so also no flag to enable it.
Alternative would be mimicking findall in Python by repeatedly calling re.search on slices of the input string, advancing each slice after a match

MICROPYPATH is the equivalent. Would make no sense to use PYTHONPATH because that would make it very awkward to use both Python and MicroPython.

I don't think any 'bootstrap script' or similar is implemented for PC ports.