In applications with multiple threads, the class variables in RRD::Wrapper storing pointers were sporadically causing segfaults since they could be freed multiple times or freed while in use by another thread. Not sure if you'd rather refactor another way, but using thread-local storage seemed to be the least intrusive fix.
In applications with multiple threads, the class variables in RRD::Wrapper storing pointers were sporadically causing segfaults since they could be freed multiple times or freed while in use by another thread. Not sure if you'd rather refactor another way, but using thread-local storage seemed to be the least intrusive fix.