On 04/10/13 12:20, Rolf Eike Beer wrote: > Just some thoughts: > -some compilers may error out because you dereference an uninitialized > pointer > -some compilers may complain because you perform an operation that does > nothing, maybe prepend it with a (void) to mark that as intentional
This is the test that is currently executed, I just added some tests... > What about this: > > struct struct_with_member tmp, *tmpp; > tmp.foo = 0; > tmpp = &tmp; > return tmpp->foo; I saw at leas one case where CheckStructHasMember is used to check if a struct has a method, so "foo()" instead of "foo"... With such a change "tmp.foo() = 0" will fail, so it looks like a non compatible change to me. Also this requires that the struct has a public constructor. If you are not against it, I would also like to change "int main()" to "int main(int argc, char **argv)" because I found a library (ACE [1]) that causes the test to fail on windows due to some weird define (see [2]) [1]http://www.cs.wustl.edu/~schmidt/ACE.html [2]http://doxygen.theaceorb.nl/libace-doc/a01553_source.html So what about something like this? int main(int argc, char **argv) { (void)(argc); // To avoid warnings for unused parameters (void)(argv); struct struct_with_member *tmp = 0; void *tmpv = (void*)&(tmp->foo); (void)(tmpv); return tmpv; } Cheers, Daniele -- Powered by www.kitware.com Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html Please keep messages on-topic and check the CMake FAQ at: http://www.cmake.org/Wiki/CMake_FAQ Follow this link to subscribe/unsubscribe: http://public.kitware.com/cgi-bin/mailman/listinfo/cmake-developers
