This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
Other format: | [Raw text] |
-Wnon-virtual-dtor isn't always what you want, defining a polymorphic object without a virtual destructor is not necessarily a mistake. You may never delete such an object so instead of warning when the class is defined it's more useful to warn only when the class is deleted, as Clang does with -Wdelete-non-virtual-dtor This patch implements the same warning for G++. Unlike Clang's, this diagnostic distinguishes the case where the pointer might refer to a base class from the case where it definitely does because the static type is abstract. There's no warning for a C++0x final class, since the pointer must refer to the most-derived type. I'm not entirely happy with the diagnostic text as I don't think we usually refer to undefined behaviour. Better suggestions welcome. The new test uses -std=gnu++0x in order to test a 'final' class, should that be split into a separate c++0x test, so the rest uses gnu++98? I'm not sure why PR c++/7302 is still open, but I think this would allow it to be closed, as it issues a warning for the delete expression in comment 25. Bootstrapped and tested x86_64-linux OK for trunk? c-family/ChangeLog: * c.opt: Add -Wdelete-non-virtual-dtor. * c-opts.c (c_common_handle_option): Include it in -Wall. cp/ChangeLog: * init.c (build_delete): Warn when deleting type with non-virtual destructor. testsuite/ChangeLog: * testsuite/g++.dg/warn/delete-non-virtual-dtor.cc: New. ChangeLog: * doc/invoke.texi: Document -Wdelete-non-virtual-dtor.
Attachment:
delnonvdtor.txt
Description: Text document
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |