libstdc++-v3/doc/html/manual/bk01pt04ch11.html

   1 <?xml version="1.0" encoding="UTF-8" standalone="no"?>
   2 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
   3 <html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Chapter 11. Memory</title><meta name="generator" content="DocBook XSL Stylesheets V1.73.2" /><meta name="keywords" content="&#10;      ISO C++&#10;    , &#10;      library&#10;    " /><link rel="start" href="../spine.html" title="The GNU C++ Library Documentation" /><link rel="up" href="utilities.html" title="Part IV. Utilities" /><link rel="prev" href="bk01pt04ch10.html" title="Chapter 10. Pairs" /><link rel="next" href="auto_ptr.html" title="auto_ptr" /></head><body><div class="navheader"><table width="100%" summary="Navigation header"><tr><th colspan="3" align="center">Chapter 11. Memory</th></tr><tr><td width="20%" align="left"><a accesskey="p" href="bk01pt04ch10.html">Prev</a> </td><th width="60%" align="center">Part IV. Utilities</th><td width="20%" align="right"> <a accesskey="n" href="auto_ptr.html">Next</a></td></tr></table><hr /></div><div class="chapter" lang="en" xml:lang="en"><div class="titlepage"><div><div><h2 class="title"><a id="manual.util.memory"></a>Chapter 11. Memory</h2></div></div></div><div class="toc"><p><b>Table of Contents</b></p><dl><dt><span class="sect1"><a href="bk01pt04ch11.html#manual.util.memory.allocator">Allocators</a></span></dt><dd><dl><dt><span class="sect2"><a href="bk01pt04ch11.html#allocator.req">Requirements</a></span></dt><dt><span class="sect2"><a href="bk01pt04ch11.html#allocator.design_issues">Design Issues</a></span></dt><dt><span class="sect2"><a href="bk01pt04ch11.html#allocator.impl">Implementation</a></span></dt><dt><span class="sect2"><a href="bk01pt04ch11.html#allocator.using">Using a Specific Allocator</a></span></dt><dt><span class="sect2"><a href="bk01pt04ch11.html#allocator.custom">Custom Allocators</a></span></dt><dt><span class="sect2"><a href="bk01pt04ch11.html#allocator.ext">Extension Allocators</a></span></dt></dl></dd><dt><span class="sect1"><a href="auto_ptr.html">auto_ptr</a></span></dt><dd><dl><dt><span class="sect2"><a href="auto_ptr.html#auto_ptr.limitations">Limitations</a></span></dt><dt><span class="sect2"><a href="auto_ptr.html#auto_ptr.using">Use in Containers</a></span></dt></dl></dd><dt><span class="sect1"><a href="shared_ptr.html">shared_ptr</a></span></dt><dd><dl><dt><span class="sect2"><a href="shared_ptr.html#shared_ptr.req">Requirements</a></span></dt><dt><span class="sect2"><a href="shared_ptr.html#shared_ptr.design_issues">Design Issues</a></span></dt><dt><span class="sect2"><a href="shared_ptr.html#shared_ptr.impl">Implementation</a></span></dt><dt><span class="sect2"><a href="shared_ptr.html#shared_ptr.using">Use</a></span></dt><dt><span class="sect2"><a href="shared_ptr.html#shared_ptr.ack">Acknowledgments</a></span></dt></dl></dd></dl></div><p>
   4     Memory contains three general areas. First, function and operator
   5     calls via <code class="function">new</code> and <code class="function">delete</code>
   6     operator or member function calls.  Second, allocation via
   7     <code class="classname">allocator</code>. And finally, smart pointer and
   8     intelligent pointer abstractions.
   9   </p><div class="sect1" lang="en" xml:lang="en"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a id="manual.util.memory.allocator"></a>Allocators</h2></div></div></div><p>
  10  Memory management for Standard Library entities is encapsulated in a
  11  class template called <code class="classname">allocator</code>. The
  12  <code class="classname">allocator</code> abstraction is used throughout the
  13  library in <code class="classname">string</code>, container classes,
  14  algorithms, and parts of iostreams. This class, and base classes of
  15  it, are the superset of available free store (“<span class="quote">heap</span>”)
  16  management classes.
  17 </p><div class="sect2" lang="en" xml:lang="en"><div class="titlepage"><div><div><h3 class="title"><a id="allocator.req"></a>Requirements</h3></div></div></div><p>
  18     The C++ standard only gives a few directives in this area:
  19   </p><div class="itemizedlist"><ul type="disc"><li><p>
  20        When you add elements to a container, and the container must
  21        allocate more memory to hold them, the container makes the
  22        request via its <span class="type">Allocator</span> template
  23        parameter, which is usually aliased to
  24        <span class="type">allocator_type</span>.  This includes adding chars
  25        to the string class, which acts as a regular STL container in
  26        this respect.
  27       </p></li><li><p>
  28        The default <span class="type">Allocator</span> argument of every
  29        container-of-T is <code class="classname">allocator&lt;T&gt;</code>.
  30        </p></li><li><p>
  31        The interface of the <code class="classname">allocator&lt;T&gt;</code> class is
  32          extremely simple.  It has about 20 public declarations (nested
  33          typedefs, member functions, etc), but the two which concern us most
  34          are:
  35        </p><pre class="programlisting">
  36          T*    allocate   (size_type n, const void* hint = 0);
  37          void  deallocate (T* p, size_type n);
  38        </pre><p>
  39          The <code class="varname">n</code> arguments in both those
  40          functions is a <span class="emphasis"><em>count</em></span> of the number of
  41          <span class="type">T</span>'s to allocate space for, <span class="emphasis"><em>not their
  42          total size</em></span>.
  43          (This is a simplification; the real signatures use nested typedefs.)
  44        </p></li><li><p>
  45          The storage is obtained by calling <code class="function">::operator
  46          new</code>, but it is unspecified when or how
  47          often this function is called.  The use of the
  48          <code class="varname">hint</code> is unspecified, but intended as an
  49          aid to locality if an implementation so
  50          desires. <code class="constant">[20.4.1.1]/6</code>
  51        </p></li></ul></div><p>
  52      Complete details cam be found in the C++ standard, look in
  53      <code class="constant">[20.4 Memory]</code>.
  54    </p></div><div class="sect2" lang="en" xml:lang="en"><div class="titlepage"><div><div><h3 class="title"><a id="allocator.design_issues"></a>Design Issues</h3></div></div></div><p>
  55     The easiest way of fulfilling the requirements is to call
  56     <code class="function">operator new</code> each time a container needs
  57     memory, and to call <code class="function">operator delete</code> each time
  58     the container releases memory. This method may be <a class="ulink" href="http://gcc.gnu.org/ml/libstdc++/2001-05/msg00105.html" target="_top">slower</a>
  59     than caching the allocations and re-using previously-allocated
  60     memory, but has the advantage of working correctly across a wide
  61     variety of hardware and operating systems, including large
  62     clusters. The <code class="classname">__gnu_cxx::new_allocator</code>
  63     implements the simple operator new and operator delete semantics,
  64     while <code class="classname">__gnu_cxx::malloc_allocator</code>
  65     implements much the same thing, only with the C language functions
  66     <code class="function">std::malloc</code> and <code class="function">free</code>.
  67   </p><p>
  68     Another approach is to use intelligence within the allocator
  69     class to cache allocations. This extra machinery can take a variety
  70     of forms: a bitmap index, an index into an exponentially increasing
  71     power-of-two-sized buckets, or simpler fixed-size pooling cache.
  72     The cache is shared among all the containers in the program: when
  73     your program's <code class="classname">std::vector&lt;int&gt;</code> gets
  74   cut in half and frees a bunch of its storage, that memory can be
  75   reused by the private
  76   <code class="classname">std::list&lt;WonkyWidget&gt;</code> brought in from
  77   a KDE library that you linked against.  And operators
  78   <code class="function">new</code> and <code class="function">delete</code> are not
  79   always called to pass the memory on, either, which is a speed
  80   bonus. Examples of allocators that use these techniques are
  81   <code class="classname">__gnu_cxx::bitmap_allocator</code>,
  82   <code class="classname">__gnu_cxx::pool_allocator</code>, and
  83   <code class="classname">__gnu_cxx::__mt_alloc</code>.
  84   </p><p>
  85     Depending on the implementation techniques used, the underlying
  86     operating system, and compilation environment, scaling caching
  87     allocators can be tricky. In particular, order-of-destruction and
  88     order-of-creation for memory pools may be difficult to pin down
  89     with certainty, which may create problems when used with plugins
  90     or loading and unloading shared objects in memory. As such, using
  91     caching allocators on systems that do not support
  92     <code class="function">abi::__cxa_atexit</code> is not recommended.
  93   </p></div><div class="sect2" lang="en" xml:lang="en"><div class="titlepage"><div><div><h3 class="title"><a id="allocator.impl"></a>Implementation</h3></div></div></div><div class="sect3" lang="en" xml:lang="en"><div class="titlepage"><div><div><h4 class="title"><a id="id425964"></a>Interface Design</h4></div></div></div><p>
  94      The only allocator interface that
  95      is support is the standard C++ interface. As such, all STL
  96      containers have been adjusted, and all external allocators have
  97      been modified to support this change.
  98    </p><p>
  99      The class <code class="classname">allocator</code> just has typedef,
 100    constructor, and rebind members. It inherits from one of the
 101    high-speed extension allocators, covered below. Thus, all
 102    allocation and deallocation depends on the base class.
 103    </p><p>
 104      The base class that <code class="classname">allocator</code> is derived from
 105      may not be user-configurable.
 106 </p></div><div class="sect3" lang="en" xml:lang="en"><div class="titlepage"><div><div><h4 class="title"><a id="id424963"></a>Selecting Default Allocation Policy</h4></div></div></div><p>
 107      It's difficult to pick an allocation strategy that will provide
 108    maximum utility, without excessively penalizing some behavior. In
 109    fact, it's difficult just deciding which typical actions to measure
 110    for speed.
 111    </p><p>
 112      Three synthetic benchmarks have been created that provide data
 113      that is used to compare different C++ allocators. These tests are:
 114    </p><div class="orderedlist"><ol type="1"><li><p>
 115        Insertion.
 116        </p><p>
 117        Over multiple iterations, various STL container
 118      objects have elements inserted to some maximum amount. A variety
 119      of allocators are tested.
 120      Test source for <a class="ulink" href="http://gcc.gnu.org/viewcvs/trunk/libstdc%2B%2B-v3/testsuite/performance/23_containers/insert/sequence.cc?view=markup" target="_top">sequence</a>
 121      and <a class="ulink" href="http://gcc.gnu.org/viewcvs/trunk/libstdc%2B%2B-v3/testsuite/performance/23_containers/insert/associative.cc?view=markup" target="_top">associative</a>
 122      containers.
 123        </p></li><li><p>
 124        Insertion and erasure in a multi-threaded environment.
 125        </p><p>
 126        This test shows the ability of the allocator to reclaim memory
 127      on a pre-thread basis, as well as measuring thread contention
 128      for memory resources.
 129      Test source
 130     <a class="ulink" href="http://gcc.gnu.org/viewcvs/trunk/libstdc%2B%2B-v3/testsuite/performance/23_containers/insert_erase/associative.cc?view=markup" target="_top">here</a>.
 131        </p></li><li><p>
 132          A threaded producer/consumer model.
 133        </p><p>
 134        Test source for
 135      <a class="ulink" href="http://gcc.gnu.org/viewcvs/trunk/libstdc%2B%2B-v3/testsuite/performance/23_containers/producer_consumer/sequence.cc?view=markup" target="_top">sequence</a>
 136      and
 137      <a class="ulink" href="http://gcc.gnu.org/viewcvs/trunk/libstdc%2B%2B-v3/testsuite/performance/23_containers/producer_consumer/associative.cc?view=markup" target="_top">associative</a>
 138      containers.
 139      </p></li></ol></div><p>
 140      The current default choice for
 141      <code class="classname">allocator</code> is
 142      <code class="classname">__gnu_cxx::new_allocator</code>.
 143    </p></div><div class="sect3" lang="en" xml:lang="en"><div class="titlepage"><div><div><h4 class="title"><a id="id458581"></a>Disabling Memory Caching</h4></div></div></div><p>
 144       In use, <code class="classname">allocator</code> may allocate and
 145       deallocate using implementation-specified strategies and
 146       heuristics. Because of this, every call to an allocator object's
 147       <code class="function">allocate</code> member function may not actually
 148       call the global operator new. This situation is also duplicated
 149       for calls to the <code class="function">deallocate</code> member
 150       function.
 151     </p><p>
 152      This can be confusing.
 153    </p><p>
 154      In particular, this can make debugging memory errors more
 155      difficult, especially when using third party tools like valgrind or
 156      debug versions of <code class="function">new</code>.
 157    </p><p>
 158      There are various ways to solve this problem. One would be to use
 159      a custom allocator that just called operators
 160      <code class="function">new</code> and <code class="function">delete</code>
 161      directly, for every allocation. (See
 162      <code class="filename">include/ext/new_allocator.h</code>, for instance.)
 163      However, that option would involve changing source code to use
 164      a non-default allocator. Another option is to force the
 165      default allocator to remove caching and pools, and to directly
 166      allocate with every call of <code class="function">allocate</code> and
 167      directly deallocate with every call of
 168      <code class="function">deallocate</code>, regardless of efficiency. As it
 169      turns out, this last option is also available.
 170    </p><p>
 171      To globally disable memory caching within the library for the
 172      default allocator, merely set
 173      <code class="constant">GLIBCXX_FORCE_NEW</code> (with any value) in the
 174      system's environment before running the program. If your program
 175      crashes with <code class="constant">GLIBCXX_FORCE_NEW</code> in the
 176      environment, it likely means that you linked against objects
 177      built against the older library (objects which might still using the
 178      cached allocations...).
 179   </p></div></div><div class="sect2" lang="en" xml:lang="en"><div class="titlepage"><div><div><h3 class="title"><a id="allocator.using"></a>Using a Specific Allocator</h3></div></div></div><p>
 180      You can specify different memory management schemes on a
 181      per-container basis, by overriding the default
 182      <span class="type">Allocator</span> template parameter.  For example, an easy
 183       (but non-portable) method of specifying that only <code class="function">malloc</code> or <code class="function">free</code>
 184       should be used instead of the default node allocator is:
 185    </p><pre class="programlisting">
 186     std::list &lt;int, __gnu_cxx::malloc_allocator&lt;int&gt; &gt;  malloc_list;</pre><p>
 187       Likewise, a debugging form of whichever allocator is currently in use:
 188     </p><pre class="programlisting">
 189     std::deque &lt;int, __gnu_cxx::debug_allocator&lt;std::allocator&lt;int&gt; &gt; &gt;  debug_deque;
 190       </pre></div><div class="sect2" lang="en" xml:lang="en"><div class="titlepage"><div><div><h3 class="title"><a id="allocator.custom"></a>Custom Allocators</h3></div></div></div><p>
 191     Writing a portable C++ allocator would dictate that the interface
 192     would look much like the one specified for
 193     <code class="classname">allocator</code>. Additional member functions, but
 194     not subtractions, would be permissible.
 195   </p><p>
 196      Probably the best place to start would be to copy one of the
 197    extension allocators: say a simple one like
 198    <code class="classname">new_allocator</code>.
 199    </p></div><div class="sect2" lang="en" xml:lang="en"><div class="titlepage"><div><div><h3 class="title"><a id="allocator.ext"></a>Extension Allocators</h3></div></div></div><p>
 200     Several other allocators are provided as part of this
 201     implementation.  The location of the extension allocators and their
 202     names have changed, but in all cases, functionality is
 203     equivalent. Starting with gcc-3.4, all extension allocators are
 204     standard style. Before this point, SGI style was the norm. Because of
 205     this, the number of template arguments also changed. Here's a simple
 206     chart to track the changes.
 207   </p><p>
 208     More details on each of these extension allocators follows.
 209   </p><div class="orderedlist"><ol type="1"><li><p>
 210        <code class="classname">new_allocator</code>
 211        </p><p>
 212          Simply wraps <code class="function">::operator new</code>
 213          and <code class="function">::operator delete</code>.
 214        </p></li><li><p>
 215        <code class="classname">malloc_allocator</code>
 216        </p><p>
 217          Simply wraps <code class="function">malloc</code> and
 218          <code class="function">free</code>. There is also a hook for an
 219          out-of-memory handler (for
 220          <code class="function">new</code>/<code class="function">delete</code> this is
 221          taken care of elsewhere).
 222        </p></li><li><p>
 223        <code class="classname">array_allocator</code>
 224        </p><p>
 225          Allows allocations of known and fixed sizes using existing
 226          global or external storage allocated via construction of
 227          <code class="classname">std::tr1::array</code> objects. By using this
 228          allocator, fixed size containers (including
 229          <code class="classname">std::string</code>) can be used without
 230          instances calling <code class="function">::operator new</code> and
 231          <code class="function">::operator delete</code>. This capability
 232          allows the use of STL abstractions without runtime
 233          complications or overhead, even in situations such as program
 234          startup. For usage examples, please consult the testsuite.
 235        </p></li><li><p>
 236        <code class="classname">debug_allocator</code>
 237        </p><p>
 238          A wrapper around an arbitrary allocator A.  It passes on
 239          slightly increased size requests to A, and uses the extra
 240          memory to store size information.  When a pointer is passed
 241          to <code class="function">deallocate()</code>, the stored size is
 242          checked, and <code class="function">assert()</code> is used to
 243          guarantee they match.
 244        </p></li><li><p>
 245         <code class="classname">throw_allocator</code>
 246         </p><p>
 247           Includes memory tracking and marking abilities as well as hooks for
 248           throwing exceptions at configurable intervals (including random,
 249           all, none).
 250         </p></li><li><p>
 251        <code class="classname">__pool_alloc</code>
 252        </p><p>
 253          A high-performance, single pool allocator.  The reusable
 254          memory is shared among identical instantiations of this type.
 255          It calls through <code class="function">::operator new</code> to
 256          obtain new memory when its lists run out.  If a client
 257          container requests a block larger than a certain threshold
 258          size, then the pool is bypassed, and the allocate/deallocate
 259          request is passed to <code class="function">::operator new</code>
 260          directly.
 261        </p><p>
 262          Older versions of this class take a boolean template
 263          parameter, called <code class="varname">thr</code>, and an integer template
 264          parameter, called <code class="varname">inst</code>.
 265        </p><p>
 266          The <code class="varname">inst</code> number is used to track additional memory
 267       pools.  The point of the number is to allow multiple
 268       instantiations of the classes without changing the semantics at
 269       all.  All three of
 270        </p><pre class="programlisting">
 271     typedef  __pool_alloc&lt;true,0&gt;    normal;
 272     typedef  __pool_alloc&lt;true,1&gt;    private;
 273     typedef  __pool_alloc&lt;true,42&gt;   also_private;
 274    </pre><p>
 275      behave exactly the same way.  However, the memory pool for each type
 276       (and remember that different instantiations result in different types)
 277       remains separate.
 278    </p><p>
 279      The library uses <span class="emphasis"><em>0</em></span> in all its instantiations.  If you
 280       wish to keep separate free lists for a particular purpose, use a
 281       different number.
 282    </p><p>The <code class="varname">thr</code> boolean determines whether the
 283    pool should be manipulated atomically or not.  When
 284    <code class="varname">thr</code> = <code class="constant">true</code>, the allocator
 285    is is thread-safe, while <code class="varname">thr</code> =
 286    <code class="constant">false</code>, and is slightly faster but unsafe for
 287    multiple threads.
 288    </p><p>
 289      For thread-enabled configurations, the pool is locked with a
 290      single big lock. In some situations, this implementation detail
 291      may result in severe performance degradation.
 292    </p><p>
 293      (Note that the GCC thread abstraction layer allows us to provide
 294      safe zero-overhead stubs for the threading routines, if threads
 295      were disabled at configuration time.)
 296    </p></li><li><p>
 297        <code class="classname">__mt_alloc</code>
 298        </p><p>
 299          A high-performance fixed-size allocator with
 300          exponentially-increasing allocations. It has its own
 301          documentation, found <a class="ulink" href="../ext/mt_allocator.html" target="_top">here</a>.
 302        </p></li><li><p>
 303        <code class="classname">bitmap_allocator</code>
 304        </p><p>
 305          A high-performance allocator that uses a bit-map to keep track
 306          of the used and unused memory locations. It has its own
 307          documentation, found <a class="ulink" href="../ext/ballocator_doc.html" target="_top">here</a>.
 308        </p></li></ol></div></div><div class="bibliography"><div class="titlepage"><div><div><h3 class="title"><a id="allocator.biblio"></a>Bibliography</h3></div></div></div><div class="biblioentry"><a id="id457599"></a><p><span class="title"><i>
 309     ISO/IEC 14882:1998 Programming languages - C++
 310     </i>. </span>
 311       isoc++_1998
 312     <span class="pagenums">20.4 Memory. </span></p></div><div class="biblioentry"><a id="id457613"></a><p><span class="title"><i>The Standard Librarian: What Are Allocators Good
 313     </i>. </span>
 314       austernm
 315     <span class="author"><span class="firstname">Matt</span> <span class="surname">Austern</span>. </span><span class="publisher"><span class="publishername">
 316         C/C++ Users Journal
 317       . </span></span><span class="biblioid">
 318       <a class="ulink" href="http://www.cuj.com/documents/s=8000/cujcexp1812austern/" target="_top">
 319       </a>
 320     . </span></p></div><div class="biblioentry"><a id="id458630"></a><p><span class="title"><i>The Hoard Memory Allocator</i>. </span>
 321       emeryb
 322     <span class="author"><span class="firstname">Emery</span> <span class="surname">Berger</span>. </span><span class="biblioid">
 323       <a class="ulink" href="http://www.cs.umass.edu/~emery/hoard/" target="_top">
 324       </a>
 325     . </span></p></div><div class="biblioentry"><a id="id459380"></a><p><span class="title"><i>Reconsidering Custom Memory Allocation</i>. </span>
 326       bergerzorn
 327     <span class="author"><span class="firstname">Emery</span> <span class="surname">Berger</span>. </span><span class="author"><span class="firstname">Ben</span> <span class="surname">Zorn</span>. </span><span class="author"><span class="firstname">Kathryn</span> <span class="surname">McKinley</span>. </span><span class="copyright">Copyright © 2002 OOPSLA. </span><span class="biblioid">
 328       <a class="ulink" href="http://www.cs.umass.edu/~emery/pubs/berger-oopsla2002.pdf" target="_top">
 329       </a>
 330     . </span></p></div><div class="biblioentry"><a id="id470488"></a><p><span class="title"><i>Allocator Types</i>. </span>
 331       kreftlanger
 332     <span class="author"><span class="firstname">Klaus</span> <span class="surname">Kreft</span>. </span><span class="author"><span class="firstname">Angelika</span> <span class="surname">Langer</span>. </span><span class="publisher"><span class="publishername">
 333         C/C++ Users Journal
 334       . </span></span><span class="biblioid">
 335       <a class="ulink" href="http://www.langer.camelot.de/Articles/C++Report/Allocators/Allocators.html" target="_top">
 336       </a>
 337     . </span></p></div><div class="biblioentry"><a id="id454967"></a><p><span class="title"><i>The C++ Programming Language</i>. </span>
 338       tcpl
 339     <span class="author"><span class="firstname">Bjarne</span> <span class="surname">Stroustrup</span>. </span><span class="copyright">Copyright © 2000 . </span><span class="pagenums">19.4 Allocators. </span><span class="publisher"><span class="publishername">
 340         Addison Wesley
 341       . </span></span></p></div><div class="biblioentry"><a id="id453321"></a><p><span class="title"><i>Yalloc: A Recycling C++ Allocator</i>. </span>
 342       yenf
 343     <span class="author"><span class="firstname">Felix</span> <span class="surname">Yen</span>. </span><span class="copyright">Copyright ©  . </span><span class="biblioid">
 344       <a class="ulink" href="http://home.earthlink.net/~brimar/yalloc/" target="_top">
 345       </a>
 346     . </span></p></div></div></div></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="bk01pt04ch10.html">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="utilities.html">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="auto_ptr.html">Next</a></td></tr><tr><td width="40%" align="left" valign="top">Chapter 10. Pairs </td><td width="20%" align="center"><a accesskey="h" href="../spine.html">Home</a></td><td width="40%" align="right" valign="top"> auto_ptr</td></tr></table></div></body></html>