haoNoQ wrote: Yes I think you may be able to reclaim some true positives by instead asking if the entire call stack, from the constructor/destructor down to the virtual call and everything in between, is in system headers.
But I'm not sure if it's worth the risk of running into more stuff where technically some frames are in user code but it's still not the users fault. And in this checker that's still quite likely. (If you could edit boost headers you'd probably solve this by marking some classes and/or their shutdown methods `final` right? So like, even then, it wouldn't be a real error, it's effectively just a suppression, and the warning is still a false alarm because *in reality* there are *currently* no overrides?) https://github.com/llvm/llvm-project/pull/184183 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
