On Thu, Nov 22, 2012 at 9:18 AM, Howard Hinnant <span dir="ltr"><<a href="mailto:hhinnant@apple.com" target="_blank">hhinnant@apple.com</a>></span> wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class="HOEnZb"><div class="h5">On Nov 22, 2012, at 11:27 AM, Ryan Molden <<a href="mailto:ryanmolden@gmail.com">ryanmolden@gmail.com</a>> wrote:<br>
<br>
> On Thu, Nov 22, 2012 at 2:14 AM, Richard Smith <<a href="mailto:richard@metafoo.co.uk">richard@metafoo.co.uk</a>> wrote:<br>
> On Wed, Nov 21, 2012 at 9:52 PM, Ryan Molden <<a href="mailto:ryanmolden@gmail.com">ryanmolden@gmail.com</a>> wrote:<br>
> > Is this the right alias for discussion of Clang implementation specifics? Or<br>
> > would that be better over on cfe-commits?<br>
><br>
> This is the right place.<br>
><br>
> > Assuming this is an appropriate alias (or that I can simply forward this on<br>
> > if not), I am fixing some missing compiler intrinsics that Clang doesn't<br>
> > provide but the VS 2012 STL headers require for some of their type traits<br>
> > implementations.<br>
> ><br>
> > In doing this I have found a couple of things in Clang's current type trait<br>
> > support that seem incorrect by my reading of the standards, but I wanted to<br>
> > run this by other people before I changed anything.<br>
> ><br>
> > The first issue is this code:<br>
> ><br>
> > struct NotTrivial {<br>
> > NotTrivial() = delete;<br>
> > };<br>
> ><br>
> > int main(int,char**)<br>
> > {<br>
> > static_assert(!__is_trivial(NotTrivial), "A class with no default<br>
> > constructor doesn't meet the definition of a trivial class");<br>
> > return 0;<br>
> > }<br>
> ><br>
> > This fires today in Clang trunk.<br>
> ><br>
> > Looking at the standard I see this:<br>
> ><br>
> > The definition of a trivial type is n3376 3.9 [basic.types]/9<br>
> ><br>
> > Arithmetic types (3.9.1), enumeration types, pointer types, pointer to<br>
> > member types (3.9.2), std::nullptr_t, and cv-qualified<br>
> > versions of these types (3.9.3) are collectively called scalar types. Scalar<br>
> > types, POD classes (Clause 9), arrays of such types and cv-qualified<br>
> > versions of these types (3.9.3) are collectively called POD types. Scalar<br>
> > types, trivially copyable class types (Clause 9), arrays of such types, and<br>
> > cv-qualified versions of these types (3.9.3) are collectively called<br>
> > trivially copyable types. Scalar types, trivial class types (Clause 9),<br>
> > arrays of such types and cv-qualified versions of these types (3.9.3) are<br>
> > collectively called trivial types. Scalar types, standard-layout class types<br>
> > (Clause 9), arrays of such types and cv-qualified<br>
> > versions of these types (3.9.3) are collectively called standard layout<br>
> > types.<br>
> ><br>
> > The referenced Clause 9 says (n3376 9 [class]/6):<br>
> ><br>
> > A trivial class is a class that has a trivial default constructor (12.1) and<br>
> > is trivially copyable.<br>
> > [ Note: In particular, a trivially copyable or trivial class does not have<br>
> > virtual functions or virtual base<br>
> ><br>
> > Since NotTrivial has no default constructor it seems it can't meet this<br>
> > definition.<br>
><br>
> Clang is behaving correctly. NotTrivial has a default constructor,<br>
> which is deleted, and thus not user-provided. It meets the other<br>
> constraints for triviality, so it is a trivial default constructor.<br>
><br>
> (Note that deleted functions still exist, and can still be trivial.)<br>
><br>
> > The second issue (this one I am less sure about) is in this code (this also<br>
> > fires in Clang trunk):<br>
> ><br>
> > struct NotTriviallyCopyable {<br>
> > NotTriviallyCopyable(const NotTriviallyCopyable&) = delete;<br>
> > NotTriviallyCopyable& operator=(const NotTriviallyCopyable&) = delete;<br>
> > };<br>
> ><br>
> > int main(int,char**)<br>
> > {<br>
> > static_assert(!__is_trivially_copyable(NotTriviallyCopyable), "A class<br>
> > with no copy constructor or copy assignment operator is not trivially<br>
> > copyable");<br>
> > return 0;<br>
> > }<br>
> ><br>
> > This one hinges on the definition of trivially-copyable, which seems like it<br>
> > can be read two ways:<br>
> ><br>
> > n3376 9 [class]/6:<br>
> ><br>
> > A trivially copyable class is a class that:<br>
> > — has no non-trivial copy constructors (12.8),<br>
> > — has no non-trivial move constructors (12.8),<br>
> > — has no non-trivial copy assignment operators (13.5.3, 12.8),<br>
> > — has no non-trivial move assignment operators (13.5.3, 12.8), and<br>
> > — has a trivial destructor (12.4).<br>
> ><br>
> > It seems since this class has NO copy constructors or copy assignment<br>
> > operators that it shouldn't be considered trivially copyable.<br>
><br>
> Clang is correct here too. This class has a trivial (deleted) copy<br>
> constructor and a trivial (deleted) copy assignment. It has no move<br>
> constructor and no move assignment. It also has a trivial destructor.<br>
> Therefore it is trivially copyable.<br>
><br>
> Interesting. My conceptualization of deleted functions was that they any reference to them in code would be ill-formed. It isn't clear how one uses some of the type traits (like is_default_constructible, is_copy_constructible, etc..) in meaningful ways in the face of deleted functions. It seems like it would lead one down a path of choosing functions/code paths that end up being ill-formed because the thing you tested for and were told existed by the compiler was actually deleted.<br>
><br>
> Another example, I notice that UTT_HasTrivialDefaultConstructor maps to the intrinsic __has_trivial_constructor, which reports true for a class with a deleted default constructor. Wouldn't that naturally lead one to think code like T t; is valid, when it would not be?<br>
<br>
</div></div>Some traits are more useful than others. I recommend std::is_trivially_default_constructible, std::is_trivially_copy_constructible and std::is_trivially_copy_assignable for answering the questions you are asking.<br>
<span class="HOEnZb"><font color="#888888"><br>
Howard<br>
<br>
</font></span></blockquote></div><br><div>Thanks to both of you for the clarifications.</div><div><br>Ryan</div>