Unable to compile

[holkono]

Hi, I've been trying to compile this for several days now but can't get it compile.

I have tried several versions of STM32cubeIDE, the old system workbench and IAR trail versions but can't get nothing to work.

Do you have some compile instructions or at this point I'd be happy with a precompiled binary.

Thanks.

[jankae]

Okay, lets see. First a word of warning: this is a one-off project which I did quite a while ago and may not remember all the details. The LCR still works and I occasionally use it, but I do not intend to maintain this project very well. The repo is more or less just a backup and minimal documentation for me.

Checking out the code and starting STM32CubeIDE 1.8.0 (yes, I am still on that old version) got me a couple of compilation errors:

error: partial specialization 'struct {anonymous}::pmf_cast<R (*)(F, A ...), C, p>' is not more specialized than [-fpermissive]

Looks like the newer compiler version does not like the template magic for function pointer casts (and I am not proficient enough in C++ templates to improve the code if that is possible).

Adding -fpermissive to the C++ compiler as suggested by that error message gets me a step further.

I am now stuck with a linker error:

Error: Cannot find the specified linker script. Check the linker settings in the build configuration.

Checking the linker script path reveals a strange filename:

The actual linker script is called STM32F303VETx_FLASH.ld. No idea why the IDE has this set to the wrong (non-existing) file.

Fixing the linker file setting, we can move on to the next error:

../Drivers/Board/Display/display.c:103: undefined reference to `selectRegister'
../Drivers/Board/Display/display.c:129: undefined reference to `setYStartStop'
[...] more undefined references

All these errors complain about functions defined in display.c, specifically all function with the keyword inline. A quick google search seems to indicate that the compiler does not like the inline keyword on its own (just in combination with static). Again, not sure why this worked at some point.

But the inline keyword is just a suggestion to the compiler anyway as far as I know. It is not really required, the optimization will take care of that.
So I replaced all inline occurrences with static in that file. And:

arm-none-eabi-size LCRMeter.elf
text data bss dec hex filename
153664 568 59192 213424 341b0 LCRMeter.elf
Finished building: default.size.stdout

arm-none-eabi-objdump -h -S LCRMeter.elf > "LCRMeter.list"
Finished building: LCRMeter.list

arm-none-eabi-objcopy -O ihex LCRMeter.elf "LCRMeter.hex"
Finished building: LCRMeter.hex

Looks like it finally worked. No idea whether that software will actually run, but here is the binary:
LCR.zip

[jankae]

Looks like #8 also took care of compilation errors along with other improvements. You may be better off just using that fork.

[holkono]

Wow that was fast!

Many many thanks for your instructions! I've been banging my head against the wall but my programming experience is more or less none existing and therefore don't have the skills and knowledge to understand the logs.

Thanks!