On 12 June 2017 at 09:40, Rüdiger Meier <sweet_f_a@xxxxxx> wrote: > On 06/11/2017 10:36 AM, Sami Kerola wrote: >> >> On 9 June 2017 at 22:10, Rüdiger Meier <sweet_f_a@xxxxxx> wrote: > > >>> Sorry this was broken. But I've checked now all clang and gcc versions. >>> Looks >>> like for clang the warning is only enabled when using C++ (clang++). So >>> the >>> fallthrough warning appears currently only on gcc>=7 and this works to >>> disable >>> it in a compatible way: >>> >>> #ifndef __has_cpp_attribute >>> # define __has_cpp_attribute(x) 0 >>> #endif >>> >>> #if __has_cpp_attribute(fallthrough) >>> # define UL_FALLTHROUGH __attribute__((fallthrough)) >>> #else >>> # define UL_FALLTHROUGH /* fall through */ >>> #endif >> >> >> Thank you for review Rudi. I had a look of the c.h again and did the >> change a little bit differently: >> >> #ifdef __has_cpp_attribute >> # if __has_cpp_attribute(deprecated) > > > Is this a typoo? s/deprecated/fallthrough Hi Rudi, Yes, it was a typo. Fixed in following. https://github.com/kerolasa/lelux-utiliteetit/commit/aa1ed6ac309a2bd6f0c5371a3117618074e7634a >> # define UL_FALLTHROUGH __attribute__((fallthrough)) >> # elif __has_cpp_attribute(clang::fallthrough) >> # define UL_FALLTHROUGH [[clang::fallthrough]] > > > I still wonder. Have you really seen such a fallthrough warning from clang? > I've even tried with explicit -Wimplicit-fallthrough but could not get one. > > As far as I understood these namespaced "clang::" attributes are for C++ > only. > > man clang: > Non-standard C++11 Attributes > Clang's non-standard C++11 attributes live in the clang attribute > namespace > > So IMO it makes no sense to use clang::fallthrough for C code. Removed in same commit. Strange enough I'm pretty sure that attribute worked. >> # endif >> #endif >> #ifndef UL_FALLTHROUGH >> # define UL_FALLTHROUGH /* FALLTHRU */ >> #endif > > > BTW just for cosmetics. Maybe we should rename UL_FALLTHROUGH to > __ul_fallthrough > like we have already __ul_calloc_size. IMO it would look more obvious to the > reader. I don't have strong opinion about that, changed to one you proposed. > Also I would use it without semicolon because some compilers may complain > about > useless ";" statements. Semicolon is removed. Whether this will fool various indent tools is something to be seen. -- Sami Kerola http://www.iki.fi/kerolasa/ -- To unsubscribe from this list: send the line "unsubscribe util-linux" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
