On Nov 21, 2013, at 11:02 AM, Marshall Clow <[email protected]> wrote:
> On Nov 21, 2013, at 7:34 AM, Yaron Keren <[email protected]> wrote: > >> Compiling libcxx with gcc 4.6.4 (MingW) produces these errors: >> >> type_traits:3280:31: error: expected primary-expression before 'decltype' >> type_traits:3280:29: error: expected ';' at end of member declaration >> >> memory:2415:49: error: function >> 'std::__1::default_delete<_Tp>::default_delete()' defaulted on its first >> declaration must not have an exception-specification >> memory:2435:49: error: function 'std::__1::default_delete<_Tp >> []>::default_delete()' defaulted on its first declaration must not have an >> exception-specification >> >> the attached patch defines _LIBCPP_HAS_NO_ADVANCED_SFINAE and >> _LIBCPP_HAS_NO_DEFAULTED_FUNCTIONS for gcc version < 4.7, making the library >> compile with gcc 4.6.4. >> >> Yaron > > I have no problems with the contents of this patch, but I think it’s time to > ask a different question: > > What compilers are we willing to support for libc++? > Do we care about gcc 4.6? 4.5? 4.4? 4.3? 4.2? 4.1? > Is anyone testing with these compilers? Imho we should not be supporting compilers for which no one is using libc++. Every additional compiler supported adds additional complexity to the libc++ code base. It isn't cost-free. I think a rationale for every supported compiler needs to be made. We should know why we're supporting a compiler. Howard _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
