------- Comment #11 from paolo dot carlini at oracle dot com 2010-03-22 17:07 ------- We discussed a bit the issue with Jason in Pittsburgh *before* realizing that likely the C++1x WD is wrong about not categorizing strPOD as POD, which now seems the real issue. My personal point of view is still that all the builtins should reflect, consistently, the ISO C++1x semantics. As an interim solution, until the ISO defect is resolved, we could have an __is_cxx98_pod and an __is_pod. That would be my preference. Jason - again, before realizing that we have a real ISO issue - proposed changing back __is_pod to the c++98 semantics and using __is_trivially_copyable and __is_standard_layout to implement std::is_pod. Frankly, long term, I don't think this is the most consistent and clear solution, in particular to people using the naked builtins, which then would have to learn that *all* the builtins reflect the C++1x semantics *beside* __is_pod. But what can I say, if you really think this is the best interim solution, I can live with it, only let's make **really** sure an ISO issue is opened and resolved quickly, clarifying the mess + let's document those semantics.
-- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43333
