gcc/libstdc++-v3/docs/html/20_util/howto.html

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0//EN">
<HTML>
<HEAD>
   <META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
   <META NAME="AUTHOR" CONTENT="pme@sources.redhat.com (Phil Edwards)">
   <META NAME="KEYWORDS" CONTENT="HOWTO, libstdc++, GCC, g++, libg++, STL">
   <META NAME="DESCRIPTION" CONTENT="HOWTO for the libstdc++ chapter 20.">
   <META NAME="GENERATOR" CONTENT="vi and eight fingers">
   <TITLE>libstdc++-v3 HOWTO:  Chapter 20</TITLE>
<LINK REL=StyleSheet HREF="../lib3styles.css">
<!-- $Id: howto.html,v 1.4 2000/12/03 23:47:47 jsm28 Exp $ -->
</HEAD>
<BODY>

<H1 CLASS="centered"><A NAME="top">Chapter 20:  General Utilities</A></H1>

<P>Chapter 20 deals with utility classes and functions, such as
   the oft-debated <TT>auto_ptr&lt;&gt;</TT>.
</P>


<!-- ####################################################### -->
<HR>
<H1>Contents</H1>
<UL>
   <LI><A HREF="#1"><TT>auto_ptr</TT> is not omnipotent</A>
   <LI><A HREF="#2">Automatically-generated operators</A>
   <LI><A HREF="#3">Functors</A>
   <LI><A HREF="#4">Pairs</A>
</UL>

<HR>

<!-- ####################################################### -->

<H2><A NAME="1"><TT>auto_ptr</TT> is not omnipotent</A></H2>
   <P>I'm not going to try and explain all of the fun and delicious
      things that can happen with misuse of the auto_ptr class template
      (called AP here), nor am I going to try and teach you how to use
      AP safely in the presence of copying.  The AP class is a really
      nifty idea for a smart pointer, but it is one of the dumbest of
      all the smart pointers -- and that's fine.
   </P>
   <P>AP is not meant to be a supersmart solution to all resource
      leaks everywhere.  Neither is it meant to be an effective form
      of garbage collection (although it can help, a little bit).
      And it can <EM>not</EM> be used for arrays!
   </P>
   <P>AP <EM>is</EM> meant to prevent nasty leaks in the presence of
      exceptions.  That's <EM>all</EM>.  This code is AP-friendly:
      <PRE>
    // not a recommend naming scheme, but good for web-based FAQs
    typedef std::auto_ptr&lt;MyClass&gt;  APMC;

    extern function_taking_MyClass_pointer (MyClass*);
    extern some_throwable_function ();

    void func (int data)
    {
        APMC  ap (new MyClass(data));

        some_throwable_function();   // this will throw an exception

        function_taking_MyClass_pointer (ap.get());
    }
      </PRE>When an exception gets thrown, the instance of MyClass that's
      been created on the heap will be <TT>delete</TT>'d as the stack is
      unwound past <TT>func()</TT>.
   </P>
   <P>Changing that code as follows is <EM>not</EM> AP-friendly:
      <PRE>
        APMC  ap (new MyClass[22]);
      </PRE>You will get the same problems as you would without the use
      of AP:
      <PRE>
        char*  array = new char[10];       // array new...
        ...
        delete array;                      // ...but single-object delete
      </PRE>
   </P>
   <P>AP cannot tell whether the pointer you've passed at creation points
      to one or many things.  If it points to many things, you are about
      to die.  AP is trivial to write, however, so you could write your
      own <TT>auto_array_ptr</TT> for that situation (in fact, this has
      been done many times; check the newsgroups, Usenet, Boost, etc).
   </P>
   <P>Return <A HREF="#top">to top of page</A> or
      <A HREF="../faq/index.html">to the FAQ</A>.
   </P>

<HR>
<H2><A NAME="2">Automatically-generated operators</A></H2>
   <P>Many programs (for that matter, many of the Standard algorithms
      and containers) require that you write comparison operators for
      your classes, like <TT>operator&gt;=</TT>.  As any mathmatician
      will tell you, once you have defined equality and ordering, all
      of the other comparisons are easily defined in terms of those two.
   </P>
   <P>The Committee agrees.  So, once you have written
      <TT>operator==</TT> and <TT>operator&lt;</TT> for your class
      (whether they are global or member functions is up to you), you
      can have the compiler do the grunt-work of generating the rest:
      <PRE>
    #include &lt;header_with_my_op==_and_op&lt;_defined&gt;
    #include &lt;utility&gt;
    using std::rel_ops;     // note the nested namespace!

    ...
    if ((obj1 != obj2) || (obj3 >= obj4))  foo();
      </PRE>
   </P>
   <P>Return <A HREF="#top">to top of page</A> or
      <A HREF="../faq/index.html">to the FAQ</A>.
   </P>

<HR>
<H2><A NAME="3">Functors</A></H2>
   <P>If you don't know what functors are, you're not alone.  Many people
      get slightly the wrong idea.  In the interest of not reinventing
      the wheel, we will refer you to the introduction to the functor
      concept written by SGI as part of their STL, in
      <A HREF="http://www.sgi.com/Technology/STL/functors.html">their
      http://www.sgi.com/Technology/STL/functors.html</A>.
   </P>
   <P>Return <A HREF="#top">to top of page</A> or
      <A HREF="../faq/index.html">to the FAQ</A>.
   </P>

<HR>
<H2><A NAME="4">Pairs</A></H2>
   <P>The <TT>pair&lt;T1,T2&gt;</TT> is a simple and handy way to
      carry around a pair of objects.  One is of type T1, and another of
      type T2; they may be the same type, but you don't get anything
      extra if they are.  The two members can be accessed directly, as
      <TT>.first</TT> and <TT>.second</TT>.
   </P>
   <P>Construction is simple.  The default ctor initializes each member
      with its respective default ctor.  The other simple ctor,
      <PRE>
    pair (const T1&amp; x, const T2&amp; y);
      </PRE>does what you think it does, <TT>first</TT> getting <TT>x</TT>
      and <TT>second</TT> getting <TT>y</TT>.
   </P>
   <P>There is a copy constructor, but it requires that your compiler
      handle member function templates:
      <PRE>
    template &lt;class U, class V&gt; pain (const pair&lt;U,V&gt;&amp; p);
      </PRE>The compiler will convert as necessary from U to T1 and from
      V to T2 in order to perform the respective initializations.
   </P>
   <P>The comparison operators are done for you.  Equality
      of two <TT>pair&lt;T1,T2&gt;</TT>s is defined as both <TT>first</TT>
      members comparing equal and both <TT>second</TT> members comparing
      equal; this simply delegates responsibility to the respective
      <TT>operator==</TT> functions (for types like MyClass) or builtin
      comparisons (for types like int, char, etc).
   </P>
   <P>The less-than operator is a bit odd the first time you see it.  It
      is defined as evaluating to:
      <PRE>
    x.first  &lt;  y.first  ||
        ( !(y.first  &lt;  x.first)  &amp;&amp;  x.second  &lt;  y.second )
      </PRE>
      The other operators are not defined using the <TT>rel_ops</TT>
      functions above, but their semantics are the same.
   </P>
   <P>Finally, there is a template function called <TT>make_pair</TT>
      that takes two references-to-const objects and returns an
      instance of a pair instantiated on their respective types:
      <PRE>
    pair&lt;int,MyClass&gt; p = make_pair(4,myobject);
      </PRE>
   </P>
   <P>Return <A HREF="#top">to top of page</A> or
      <A HREF="../faq/index.html">to the FAQ</A>.
   </P>


<!-- ####################################################### -->

<HR>
<P CLASS="fineprint"><EM>
Comments and suggestions are welcome, and may be sent to
<A HREF="mailto:pme@sources.redhat.com">Phil Edwards</A> or
<A HREF="mailto:gdr@gcc.gnu.org">Gabriel Dos Reis</A>.
<BR> $Id: howto.html,v 1.4 2000/12/03 23:47:47 jsm28 Exp $
</EM></P>


</BODY>
</HTML>