Fix #10431 FP one-definition-rule if struct in mutually exclusive #ifdef branches

[chrchr-github]

No description provided.

[danmar]

shouldn't you add this file odr3.cpp to whole-program_test.py in some places?

[danmar]

I guess this could become tricky.. but what happens if you have two files..

file1.cpp:

#ifdef X
struct S { int i; }; 
#endif

file2.cpp:

struct S { int i; }; 

These configurations are different but not mutually exclusive.

would it make sense to make sure that if file0 is different then all configurations are considered. If the user don't want to consider all configurations in this case then we can recommend that -D that matches compilation is used?

[chrchr-github]

shouldn't you add this file odr3.cpp to whole-program_test.py in some places?

Absolutely, I must have thought that all files would be checked automatically.

[chrchr-github]

These configurations are different but not mutually exclusive.

I have restored the warning in that case.

[chrchr-github]

Needs XML handling for --cppcheck-build-dir.

[chrchr-github]

@firewave How do I opt out of the -j2 injection?

[chrchr-github]

@firewave How do I opt out of the -j2 injection?

Guess I found it.

[firewave]

@firewave How do I opt out of the -j2 injection?

Guess I found it.

Yup. Just provide an existing -j.

Can you please file a ticket about making this work with the builddir? Thanks.

[chrchr-github]

Can you please file a ticket about making this work with the builddir? Thanks.

I don't think there is anything new here, see the existing tests:
@pytest.mark.xfail(strict=True) # no CTU without builddir