misc-include-cleaner question about general behaviour

[eospre]

I just would like to ask what the general consent about double includes in the cpp and header file of the same class is.

Lets say i have a function that returns e.g. a std::string (declared in header, defined in source file), i am currently forced to #include both, in the header and the cpp file. If i only include it in the header misc-include-cleaner will issue a warning, that no header providing "std::string" is directly included.

Now for my understanding, only including it in the header of the same class (meaning cpp and h file have the exact same name) is perfectly fine (or even preferred over double the includes).

I understand that the warning is useful for anything else (meaning not the exact case above).

Whats the general opinion on this. Might the check be adjusted in a way, that the above usecase does not issue a warning (or maybe even make it configurable)? Or is the general opinion, that even in this case, the warning is useful and we shall include necessary headers in both files?

[llvmbot]

@llvm/issue-subscribers-clang-include-cleaner

Author: Stefan Preiß (eospre)

I just would like to ask what the general consent about double includes in the cpp and header file of the same class is.

Lets say i have a function that returns e.g. a std::string (declared in header, defined in source file), i am currently forced to #include <string> both, in the header and the cpp file. If i only include it in the header misc-include-cleaner will issue a warning, that no header providing "std::string" is directly included.

Now for my understanding, only including it in the header of the same class (meaning cpp and h file have the exact same name) is perfectly fine (or even preferred over double the includes).

I understand that the warning is useful for anything else (meaning not the exact case above).

Whats the general opinion on this. Might the check be adjusted in a way, that the above usecase does not issue a warning (or maybe even make it configurable)? Or is the general opinion, that even in this case, the warning is useful and we shall include necessary headers in both files?

[EugeneZelenko]

I think this is duplicate of #122847.

[eospre]

@EugeneZelenko yes you are right. As it is marked as false-positive I guess my understanding as stated is correct and it is intended to fix it at some point, isn't it?

[EugeneZelenko]

@EugeneZelenko yes you are right. As it is marked as false-positive I guess my understanding as stated is correct and it is intended to fix it at some point, isn't it?

It's hard to tell, since looks like Clang Include Cleaner is not under active development these days. I encounter similar problems when I tried to apply Clang Include Cleaner on my work code base. See #122850 and #122852 as sources of my other griefs that you may be encountered too.