pascalj wrote: Thanks for your feedback!
> * what about "static" non const global variables Good point, forgot about these. If both are allowed, it is more about the internal linkage than it is about the namespace. I renamed the option to `AllowInternalLinkage` and permit variables in anonymous namespaces and static global variables, since they're equivalent (AFAICT). > I'm fine for having an options that control behavior of the check. But please > note that there are other aspects of "I.2: Avoid non-const global variables" > rule that anonymous namespace or static does not solve. Like race-conditions > in multithread env. Or even to force developer not to use global objects to > pass data between functions. > > Even that I see many times that warnings from this check are being nolinted, > I still think that it's doing good job. I completely agree with both you and @carlosgalvezp: this is neither an option that should be enabled by default, nor should people prefer it over nolinting without some thought. However, I see not much harm in giving someone the choice to do so. In the end, all of clang-tidy's checks are optional to some degree. Having said that: if the maintainers prefer not to have this option, this is also fine with me. https://github.com/llvm/llvm-project/pull/93827 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
