Oliver Stöneberg

reduced scope of some variables and avoided some copies

4

Another piece of preparation before applying `misc-const-correctness` fixes.

valueflow.cpp: limit `const_cast` to the necessary calls

2

This shows that there's only two calls which actually require the `const_cast` so there's no need to do it arbitrary all over the code (there was even at least one...

fixed some `modernize-use-emplace` false negatives with `std::stack` / removed `internalStlUsage` check

1

A ticket about this has been filed upstream: https://github.com/llvm/llvm-project/issues/56996.

added remaining make targets to CMake

1

With these we now have target parity between make and CMake. There's still some compiler/platform-specific code which needs to be added to CMake for full parity.

added `--check-library` to self-check

5

Related issues: ~https://trac.cppcheck.net/ticket/10105~ ~https://trac.cppcheck.net/ticket/11182~ ~https://trac.cppcheck.net/ticket/11183~ ~https://trac.cppcheck.net/ticket/11523~ ~https://trac.cppcheck.net/ticket/11543~ https://trac.cppcheck.net/ticket/11555 ~https://trac.cppcheck.net/ticket/10946~ https://trac.cppcheck.net/ticket/11313 ~https://trac.cppcheck.net/ticket/13105~ ~https://trac.cppcheck.net/ticket/13125~

An idea would be to provide a daca result file as input file. Depending on the data it could determine what to do without interaction.

This avoids the expensive C++-specific `Library::detect*()` calls in `parsedecl()` for C code. See https://trac.cppcheck.net/ticket/10663 and https://trac.cppcheck.net/ticket/11104 for more details of the underlying issues. There's possibly more parsing which could be...