Code size report:

Reference:  py/objcode: Remove `mp_obj_code_t.lnotab` field from v2 preview. [e0a9b70]Comparison: py: Add LIBS_USERMOD to LIBS. [merge of 7176acb]  mpy-cross:    +0 +0.000%    bare-arm:    +0 +0.000% minimal x86:    +0 +0.000%    unix x64:    +0 +0.000% standard      stm32:    +0 +0.000% PYBV10     mimxrt:    +0 +0.000% TEENSY40        rp2:    +0 +0.000% RPI_PICO_W       samd:    +0 +0.000% ADAFRUIT_ITSYBITSY_M4_EXPRESS  qemu rv32:    +0 +0.000% VIRT_RV32

Thanks for the patch, addingLIBS_USERMOD looks very reasonable to me.

But, the CI needs to be fixed before this PR can be merged.

If you want to quickly get theLIBS_USERMOD bit in then we can merge just that change, and not the change to usestd::vector.

Thanks! Pushed7176acb which should hopefully resolve that.

I also unlinked this PR from#18351, as it's not a full solution to get C++ modules working on all ports when actually usingstdc++.a. The CMake ports seem to build fine (I've not tested execution), but every the Make port fails to build as soon as any code is used that leveragesstdc++.a. I've added a lot of notes to#18351 about how to fix this; annoyingly, every port seems to need its own unique fixes, though there are some common fixes between them.

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.38%. Comparing base (4efc5e1) to head (dbaa315).
⚠️ Report is 12 commits behind head on master.

@@            Coverage Diff             @@##           master   #18352      +/-   ##==========================================- Coverage   98.38%   98.38%   -0.01%==========================================  Files         171      171                Lines       22297    22294       -3     ==========================================- Hits        21936    21933       -3  Misses        361      361

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report?Share it here.

https://github.com/micropython/micropython/actions/runs/19042775575/job/54383644049?pr=18352 Hmm...

LINK build-coverage/micropython/usr/bin/ld: build-coverage/cppexample/example.o:(.data.rel.local.DW.ref.__gxx_personality_v0[DW.ref.__gxx_personality_v0]+0x0): undefined reference to `__gxx_personality_v0'collect2: error: ld returned 1 exit statusmake: *** [../../py/mkrules.mk:248: build-coverage/micropython] Error 1make: Leaving directory '/home/runner/work/micropython/micropython/ports/unix'

Testing on my end withV=1, before this PR, the link command contains-lstdc++:

gcc ... -ftest-coverage -lstdc++    -Wl,-Map=...

After this PR,-lstdc++ goes away:

gcc ... -ftest-coverage     -Wl,-Map=....

Ah,unix/Makefile does not useLIBS, onlyLDFLAGS. That would explain why-lstdc++ goes missing.

Would it be acceptable to simply addLDFLAGS += $(LIBS) after the last modification ofLDFLAGS? That makes it build successfully on my end.

Would it be acceptable to simply addLDFLAGS += $(LIBS) after the last modification ofLDFLAGS? That makes it build successfully on my end.

The unix and Windows ports useLIB... and that's inconsistent with all other ports which useLIBS.

So I think the right solution here is to change the two instances ofLIB inpy/mkrules.mk andports/windows/Makefile toLIBS.

dbaa315 passes 🎉