<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
    <head>
        <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
        <title>Additional type traits: has_trivial_destructor_after_move and has_trivial_reallocation (rev1)</title>
        <style>
            #header, .note, .add {
                font-style: italic;
            }
        
            #title, .section {
                margin: 30px 0px;
            }
        
            #title, .section .name {
                font-weight: bold;
            }
            
            #title {
                font-size: 35px;
            }
            
            .section .name {
                font-size: 20px;
            }
            
            .draft {
                color: #999999;
            }
            
            .add {
                text-decoration: underline;
                color: black;
            }
            
            table {
                border-collapse: collapse;
            }
            
            tr {
                border: 1px solid black;
            }
        </style>
    </head>
    <body>
        <div id="header">
            Rodrigo Castro Campos, <a href="mailto:rcc.dark@gmail.com">rcc.dark@gmail.com</a>.<br>
            2008-08-14.<br>
            N2718=08-0228.
        </div>
        
        <div id="title">Additional type traits: has_trivial_destructor_after_move and has_trivial_reallocation (rev1)</div>
        <div class="section">
            <span class="name">0. Changes since N2698</span>
            <p>
                <ul>
                    <li>Fixed minor mistake in example and typo.
                    <li>Relaxed std::has_trivial_reallocation&lt;T&gt; default implementation to std::is_trivial&lt;T&gt; instead of std::is_pod&lt;T&gt;.
                    <li>Added wording aimed to the working draft.
                </ul>
            </p>
        </div>
        <div class="section">
            <span class="name">I. Motivation</span>
            <p>
                Many C++ programmers (myself included) would like to have destructive move semantics. 
                Even with the current proposal we can (unsafely) say almost that:
            </p>
            <pre>
    class RAII_Handle {
    public:
        // Post-condition: h_ is valid
        RAII_Handle(const char* c)
        : h_(::open(c))
        {
            if (h_ == nullptr) {
                throw false;
            }
        }
        
        RAII_Handle(RAII_Handle&& h)
        : h_(h.h_)
        {
            h.h_ = nullptr;
        }
        
        // exposition only, not self-assignment safe
        RAII_Handle& operator=(RAII_Handle&& h)
        {
            ::close(h_);
            h_ = h.h_;
            h.h_ = nullptr;
        }
        
        // if h_ is null, calling ::close( ) may crash, must protect against it
        ~RAII_Handle( )
        {
            if (h_ != nullptr) {
                ::close(h_);
            }
        }

        // if h_ is null, calling ::execute( ) may crash
        void command(const char* c)
        {
            ::execute(h_, c);
        }
        
    private:
        RAII_Handle(const RAII_Handle&);
        RAII_Handle& operator=(const RAII_Handle&);

        Handle* h_;
    };
            </pre>
            <p>
                The constructor postcondition is violated after a move. I should confess I thought 
                this was the way it should work before reading the move proposal completely (which states 
                that an object should be usable even after a move).
            </p>
            <p>
                This unsafe design may be fixed by adding open( ), close( ), and is_open( ) member functions. 
                However, if we want to use this kind of destructing move, the destructor must 
                always protect himself from dereferencing a null pointer.
            </p>
            <p>
                While a fully destructive move semantic would avoid this null-checking, the best solution 
                I have read (from what other programmers propose) would allow stack-RAII'ed variables to leak 
                resources; this is a -big no-. I fully agree with the current move proposal since its flexible 
                enough anyway and not unsafe (unless the programmer really wants it be). For heap or 
                placement-new'ed variables, however, the story may be different.
            </p>
            <p>
                In the example showed above a just-moved variable's destructor will have no side effects, in fact 
                it will just check if it must clean itself or not. This is true even for move-safe types; 
                many classes will have to be redesigned (std::string in mind, with at least one character 
                as short-string optimization) so move semantics actually allow the desirable performance boost. 
                The effects of their destructors inmediatly after a move could be ignored in most cases.
                This document proposes to type-trait this information in the same way other type traits work.
            </p>
        </div>
        <div class="section"><span class="name">II. Impact On the Standard</span>
            <p>
                Containers that separate destruction from deallocation can use this 
                information to avoid calls to the destructors if just deallocating will suffice. This concept 
                was already explained by some (see references) and I was quiet surprised not to 
                see this as part of the standard. This will give us fully-destructive-like performance without actually implementing it. 
                Implementors of containers are free to use or not this information.
            </p>
        </div>
        <div class="section"><span class="name">III. Proposed Traits and Wording</span>
            <ul>
                <li>
                    In 20.5.2 [Header &lt;type_traits&gt; synopsis [meta.type.synop]] add:
                    <pre class="draft">
template &lt;class T&gt; struct has_trivial_assign;
template &lt;class T&gt; struct has_trivial_destructor;<span class="add">
template &lt;class T&gt; struct has_trivial_destructor_after_move;
template &lt;class T&gt; struct has_trivial_reallocation;</span>
template &lt;class T&gt; struct has_nothrow_default_constructor;
template &lt;class T&gt; struct has_nothrow_copy_constructor;</pre>
                <li>
                    In 20.5.4.3 [Type properties [meta.unary.prop]] add:<br><br>
<table class="draft">
    <tr>
        <td>template &lt;class T&gt; struct has_trivial_destructor;</td>
        <td>T is a trivial type (3.9) or a reference type or a class type with a trivial destructor (12.4) or an array of such a class type.</td>
        <td>T shall be a complete type, an array of unknown bound, or (possibly cv-qualified) void.</td>
    </tr>
    <tr>
        <td class="add">template &lt;class T&gt; struct has_trivial_destructor_after_move;</td>
        <td class="add">has_trivial_destructor&lt;T&gt;::value is true or T is a class type whose destructor has no side-effects inmediatly after moving the object to be destroyed, or an array of such a class type.</td>
        <td class="add">T shall be a complete type, an array of unknown bound, or (possibly cv-qualified) void.</td>
    </tr>
    <tr>
        <td class="add">template &lt;class T&gt; struct has_trivial_reallocation;</td>
        <td class="add">T is a trivial type (3.9) or a reference type or a class type that can be reallocated in memory by bitwise-copying the object without destroying the source, or an array of such a class type.</td>
        <td class="add">T shall be a complete type, an array of unknown bound, or (possibly cv-qualified) void.</td>
    </tr>
    <tr>
        <td>template &lt;class T&gt; struct has_nothrow_default_constructor;</td>
        <td>has_trivial_default_constructor&lt;T&gt;::value is true or T is a class type with a default constructor that is known not to throw any exceptions or T is an array of such a class type.</td>
        <td>T shall be a complete type, an array of unknown bound, or (possibly cv-qualified) void.</td>
    </tr>
</table>
        </div>
        <div class="section"><span class="name">IV. References</span>
            <p>
                <ul>
                    <li>Andrei Alexandrescu (2001). Generic:Typed Buffers (I). <a href="http://www.ddj.com/cpp/184403791">http://www.ddj.com/cpp/184403791</a>
                    <li>Hinnant, Howard., Dimov, Peter., Abrahams, Dave. (2002). A Proposal to Add Move Semantics Support to the C++ Language. <a href="http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2002/n1377.htm">http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2002/n1377.htm</a>
                    <li>Paul Pedriana (2007). EASTL -- Electronic Arts Standard Template Library. <a href="http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2007/n2271.html#type_traits_extensions">http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2007/n2271.html#type_traits_extensions</a>
                </ul>
            </p>
            <p>
                Special thanks to Howard Hinnant for his feedback.
            </p>
        </div>
    </body>
</html>