xref: /netbsd-src/external/gpl3/gcc.old/dist/libstdc++-v3/doc/html/manual/iterators.html (revision 36ac495d2b3ea2b9d96377b2143ebfedac224b92)
1*36ac495dSmrg<?xml version="1.0" encoding="UTF-8" standalone="no"?>
2*36ac495dSmrg<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Chapter 10.  Iterators</title><meta name="generator" content="DocBook XSL Stylesheets Vsnapshot" /><meta name="keywords" content="ISO C++, library" /><meta name="keywords" content="ISO C++, runtime, library" /><link rel="home" href="../index.html" title="The GNU C++ Library" /><link rel="up" href="std_contents.html" title="Part II.  Standard Contents" /><link rel="prev" href="containers_and_c.html" title="Interacting with C" /><link rel="next" href="algorithms.html" title="Chapter 11.  Algorithms" /></head><body><div class="navheader"><table width="100%" summary="Navigation header"><tr><th colspan="3" align="center">Chapter 10. 
3*36ac495dSmrg  Iterators
4*36ac495dSmrg
5*36ac495dSmrg</th></tr><tr><td width="20%" align="left"><a accesskey="p" href="containers_and_c.html">Prev</a> </td><th width="60%" align="center">Part II. 
6*36ac495dSmrg    Standard Contents
7*36ac495dSmrg  </th><td width="20%" align="right"> <a accesskey="n" href="algorithms.html">Next</a></td></tr></table><hr /></div><div class="chapter"><div class="titlepage"><div><div><h2 class="title"><a id="std.iterators"></a>Chapter 10. 
8*36ac495dSmrg  Iterators
9*36ac495dSmrg  <a id="id-1.3.4.8.1.1.1" class="indexterm"></a>
10*36ac495dSmrg</h2></div></div></div><div class="toc"><p><strong>Table of Contents</strong></p><dl class="toc"><dt><span class="section"><a href="iterators.html#std.iterators.predefined">Predefined</a></span></dt><dd><dl><dt><span class="section"><a href="iterators.html#iterators.predefined.vs_pointers">Iterators vs. Pointers</a></span></dt><dt><span class="section"><a href="iterators.html#iterators.predefined.end">One Past the End</a></span></dt></dl></dd></dl></div><div class="section"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a id="std.iterators.predefined"></a>Predefined</h2></div></div></div><div class="section"><div class="titlepage"><div><div><h3 class="title"><a id="iterators.predefined.vs_pointers"></a>Iterators vs. Pointers</h3></div></div></div><p>
11*36ac495dSmrg     The following
12*36ac495dSmrgFAQ <a class="link" href="../faq.html#faq.iterator_as_pod" title="7.1.">entry</a> points out that
13*36ac495dSmrgiterators are not implemented as pointers.  They are a generalization
14*36ac495dSmrgof pointers, but they are implemented in libstdc++ as separate
15*36ac495dSmrgclasses.
16*36ac495dSmrg   </p><p>
17*36ac495dSmrg     Keeping that simple fact in mind as you design your code will
18*36ac495dSmrg      prevent a whole lot of difficult-to-understand bugs.
19*36ac495dSmrg   </p><p>
20*36ac495dSmrg     You can think of it the other way 'round, even.  Since iterators
21*36ac495dSmrg     are a generalization, that means
22*36ac495dSmrg     that <span class="emphasis"><em>pointers</em></span> are
23*36ac495dSmrg      <span class="emphasis"><em>iterators</em></span>, and that pointers can be used
24*36ac495dSmrg     whenever an iterator would be.  All those functions in the
25*36ac495dSmrg     Algorithms section of the Standard will work just as well on plain
26*36ac495dSmrg     arrays and their pointers.
27*36ac495dSmrg   </p><p>
28*36ac495dSmrg     That doesn't mean that when you pass in a pointer, it gets
29*36ac495dSmrg      wrapped into some special delegating iterator-to-pointer class
30*36ac495dSmrg      with a layer of overhead.  (If you think that's the case
31*36ac495dSmrg      anywhere, you don't understand templates to begin with...)  Oh,
32*36ac495dSmrg      no; if you pass in a pointer, then the compiler will instantiate
33*36ac495dSmrg      that template using T* as a type, and good old high-speed
34*36ac495dSmrg      pointer arithmetic as its operations, so the resulting code will
35*36ac495dSmrg      be doing exactly the same things as it would be doing if you had
36*36ac495dSmrg      hand-coded it yourself (for the 273rd time).
37*36ac495dSmrg   </p><p>
38*36ac495dSmrg     How much overhead <span class="emphasis"><em>is</em></span> there when using an
39*36ac495dSmrg      iterator class?  Very little.  Most of the layering classes
40*36ac495dSmrg      contain nothing but typedefs, and typedefs are
41*36ac495dSmrg      "meta-information" that simply tell the compiler some
42*36ac495dSmrg      nicknames; they don't create code.  That information gets passed
43*36ac495dSmrg      down through inheritance, so while the compiler has to do work
44*36ac495dSmrg      looking up all the names, your runtime code does not.  (This has
45*36ac495dSmrg      been a prime concern from the beginning.)
46*36ac495dSmrg   </p></div><div class="section"><div class="titlepage"><div><div><h3 class="title"><a id="iterators.predefined.end"></a>One Past the End</h3></div></div></div><p>This starts off sounding complicated, but is actually very easy,
47*36ac495dSmrg      especially towards the end.  Trust me.
48*36ac495dSmrg   </p><p>Beginners usually have a little trouble understand the whole
49*36ac495dSmrg      'past-the-end' thing, until they remember their early algebra classes
50*36ac495dSmrg      (see, they <span class="emphasis"><em>told</em></span> you that stuff would come in handy!) and
51*36ac495dSmrg      the concept of half-open ranges.
52*36ac495dSmrg   </p><p>First, some history, and a reminder of some of the funkier rules in
53*36ac495dSmrg      C and C++ for builtin arrays.  The following rules have always been
54*36ac495dSmrg      true for both languages:
55*36ac495dSmrg   </p><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><p>You can point anywhere in the array, <span class="emphasis"><em>or to the first element
56*36ac495dSmrg	  past the end of the array</em></span>.  A pointer that points to one
57*36ac495dSmrg	  past the end of the array is guaranteed to be as unique as a
58*36ac495dSmrg	  pointer to somewhere inside the array, so that you can compare
59*36ac495dSmrg	  such pointers safely.
60*36ac495dSmrg	</p></li><li class="listitem"><p>You can only dereference a pointer that points into an array.
61*36ac495dSmrg	  If your array pointer points outside the array -- even to just
62*36ac495dSmrg	  one past the end -- and you dereference it, Bad Things happen.
63*36ac495dSmrg	</p></li><li class="listitem"><p>Strictly speaking, simply pointing anywhere else invokes
64*36ac495dSmrg	  undefined behavior.  Most programs won't puke until such a
65*36ac495dSmrg	  pointer is actually dereferenced, but the standards leave that
66*36ac495dSmrg	  up to the platform.
67*36ac495dSmrg	</p></li></ol></div><p>The reason this past-the-end addressing was allowed is to make it
68*36ac495dSmrg      easy to write a loop to go over an entire array, e.g.,
69*36ac495dSmrg      while (*d++ = *s++);.
70*36ac495dSmrg   </p><p>So, when you think of two pointers delimiting an array, don't think
71*36ac495dSmrg      of them as indexing 0 through n-1.  Think of them as <span class="emphasis"><em>boundary
72*36ac495dSmrg      markers</em></span>:
73*36ac495dSmrg   </p><pre class="programlisting">
74*36ac495dSmrg
75*36ac495dSmrg   beginning            end
76*36ac495dSmrg     |                   |
77*36ac495dSmrg     |                   |               This is bad.  Always having to
78*36ac495dSmrg     |                   |               remember to add or subtract one.
79*36ac495dSmrg     |                   |               Off-by-one bugs very common here.
80*36ac495dSmrg     V                   V
81*36ac495dSmrg	array of N elements
82*36ac495dSmrg     |---|---|--...--|---|---|
83*36ac495dSmrg     | 0 | 1 |  ...  |N-2|N-1|
84*36ac495dSmrg     |---|---|--...--|---|---|
85*36ac495dSmrg
86*36ac495dSmrg     ^                       ^
87*36ac495dSmrg     |                       |
88*36ac495dSmrg     |                       |           This is good.  This is safe.  This
89*36ac495dSmrg     |                       |           is guaranteed to work.  Just don't
90*36ac495dSmrg     |                       |           dereference 'end'.
91*36ac495dSmrg   beginning                end
92*36ac495dSmrg
93*36ac495dSmrg   </pre><p>See?  Everything between the boundary markers is chapter of the array.
94*36ac495dSmrg      Simple.
95*36ac495dSmrg   </p><p>Now think back to your junior-high school algebra course, when you
96*36ac495dSmrg      were learning how to draw graphs.  Remember that a graph terminating
97*36ac495dSmrg      with a solid dot meant, "Everything up through this point,"
98*36ac495dSmrg      and a graph terminating with an open dot meant, "Everything up
99*36ac495dSmrg      to, but not including, this point," respectively called closed
100*36ac495dSmrg      and open ranges?  Remember how closed ranges were written with
101*36ac495dSmrg      brackets, <span class="emphasis"><em>[a,b]</em></span>, and open ranges were written with parentheses,
102*36ac495dSmrg      <span class="emphasis"><em>(a,b)</em></span>?
103*36ac495dSmrg   </p><p>The boundary markers for arrays describe a <span class="emphasis"><em>half-open range</em></span>,
104*36ac495dSmrg      starting with (and including) the first element, and ending with (but
105*36ac495dSmrg      not including) the last element:  <span class="emphasis"><em>[beginning,end)</em></span>.  See, I
106*36ac495dSmrg      told you it would be simple in the end.
107*36ac495dSmrg   </p><p>Iterators, and everything working with iterators, follows this same
108*36ac495dSmrg      time-honored tradition.  A container's <code class="code">begin()</code> method returns
109*36ac495dSmrg      an iterator referring to the first element, and its <code class="code">end()</code>
110*36ac495dSmrg      method returns a past-the-end iterator, which is guaranteed to be
111*36ac495dSmrg      unique and comparable against any other iterator pointing into the
112*36ac495dSmrg      middle of the container.
113*36ac495dSmrg   </p><p>Container constructors, container methods, and algorithms, all take
114*36ac495dSmrg      pairs of iterators describing a range of values on which to operate.
115*36ac495dSmrg      All of these ranges are half-open ranges, so you pass the beginning
116*36ac495dSmrg      iterator as the starting parameter, and the one-past-the-end iterator
117*36ac495dSmrg      as the finishing parameter.
118*36ac495dSmrg   </p><p>This generalizes very well.  You can operate on sub-ranges quite
119*36ac495dSmrg      easily this way; functions accepting a <span class="emphasis"><em>[first,last)</em></span> range
120*36ac495dSmrg      don't know or care whether they are the boundaries of an entire {array,
121*36ac495dSmrg      sequence, container, whatever}, or whether they only enclose a few
122*36ac495dSmrg      elements from the center.  This approach also makes zero-length
123*36ac495dSmrg      sequences very simple to recognize:  if the two endpoints compare
124*36ac495dSmrg      equal, then the {array, sequence, container, whatever} is empty.
125*36ac495dSmrg   </p><p>Just don't dereference <code class="code">end()</code>.
126*36ac495dSmrg   </p></div></div></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="containers_and_c.html">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="std_contents.html">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="algorithms.html">Next</a></td></tr><tr><td width="40%" align="left" valign="top">Interacting with C </td><td width="20%" align="center"><a accesskey="h" href="../index.html">Home</a></td><td width="40%" align="right" valign="top"> Chapter 11. 
127*36ac495dSmrg  Algorithms
128*36ac495dSmrg
129*36ac495dSmrg</td></tr></table></div></body></html>