On Thu, Dec 29, 2011 at 08:12:51PM -0800, Steve Kargl wrote: > The audit trail in the PR pretty much sums up the problem. > OK to commit?
Can you tune up the comments a little bit? If you look up http://gcc.gnu.org/gcc-4.7/cxx0x_status.html you'll see that most of the C++11 features are already implemented (and e.g. at_quick_exit is implemented too in some libcs), so the __cplusplus macro isn't lying. Clang apparently defines __cplusplus to 201103L much longer than GCC does, yet it supports much smaller subset of C++11. Apparently it parses the C++11 generalized attribute syntax, but at least clang 3.0 doesn't complain at all about anything between the [[ and ]] pair, [[foobar, foobaz (bar, baz, 2, 6, 0x1z2)]] void bar (void); is parsed just fine, and with [[noreturn]] it miscompiles even the first [[noreturn]] trivial testcase I wrote. See http://gcc.gnu.org/ml/gcc-patches/2011-10/msg02027.html for why we set __cplusplus to the value we do. So please leave about the "disgusting", "brain damage", "lies" etc. > 2011-12-29 Steven G. Kargl <ka...@gcc.gnu.org> > > * inclhack.def: Disgusting hack to workaround brain damage of > defining __cplusplus as 201103L with -std=c++11 when g++ does > not support c++11. > * fixincl.x: regenerated. > * genfixes: Fix to version test. > > -- > Steve > Index: inclhack.def > =================================================================== > --- inclhack.def (revision 182738) > +++ inclhack.def (working copy) > @@ -20,6 +20,21 @@ autogen definitions fixincl; > FIXINC_DEBUG = yes; > #endif > > +/* > + * g++ -std=c++11 defines __cplusplus to 201103L, which lies about > + * the level of support g++ has for the C++11 standard. > + */ Jakub
