Daniel Marjamäki

> We also require the user to be honest with us about the interface. I think that is reasonable.

Instinctively I feel that it's better with false negatives when we can't know what the function is doing. An option is to write inconclusive warnings with a separate id when...

> We have noexcept for functions that never throw, and we have [[noreturn]] for functions that always throw, but there is no attribute for functions that can throw but not...

well we could change the shebang I guess. for most users it would make better sense to use python3. For users that have old unix systems it's possible to run...

> I really thinking we should get rid of Python 2.7. I don't have a very strong opinion right now actually. Python 2.x is probably not entirely dead yet. But...

> In that case he could override it with python2 .py. We have tests that ensure we should be working with it. But the out-of-the-box behavior should be consistent i.e....

> Merge branch 'danmar:main' into main I would suggest that you use `git rebase ..` instead of `git merge ..`.

> probably needs some unit tests as well feel free to extend the tests in cppcheck/test/cfg/googletest.cpp

there are ci failures. ``` /__w/cppcheck/cppcheck/lib/checkbool.cpp:55:30: error: use nullptr [modernize-use-nullptr,-warnings-as-errors] const Token* first_var = NULL; ^~~~ nullptr /__w/cppcheck/cppcheck/lib/checkbool.cpp:56:31: error: use nullptr [modernize-use-nullptr,-warnings-as-errors] const Token* second_var = NULL; ^~~~ nullptr ```

> But you are right. It seems to be against the specs on https://clang.llvm.org/docs/JSONCompilationDatabase.html which says there MUST be a "file" field. well.. the difference seems to be by design...