Daniel Marjamäki
Daniel Marjamäki
> Why whether import a compile_commands.json or not will impact the behavior of checking the configs. Seems it is not related with that. In my humble opinion, when you import...
> In my humble opinion, when you import a compile_commands.json it is expected that we check the files and explicit configurations provided in the compile_commands.json. The Cppcheck analysis should match...
> And even though there is a compile_commands.json imported, if there is no user defined configs, we can check the extracted configs with the default max numbers of 12, just...
> Do you think we should only check the ABC1, or just run the check with an empty config define string. Actually, when getting the extracted the configs, we put...
well those CI failures needs some minor tweaks I assume?
@firewave I don't feel good about these. It is not safe. A large number of CVEs are caused by old style C buffers..
Taking a std::string would be better imho but I still feel that std::istream is better for type safety reasons.
> I also stated in the other PR that I will provide more modern interfaces but I did not have time to do that yet. I would be interested to...
I agree it's a big problem with the istream slowness..
Even though fstream is slowish my hunch is that the changes will not make a significant speedup overall in cppcheck analysis. Could you try to build cppcheck with your changes...