OpenCores
URL https://opencores.org/ocsvn/openrisc/openrisc/trunk

Subversion Repositories openrisc

[/] [openrisc/] [trunk/] [gnu-dev/] [or1k-gcc/] [libstdc++-v3/] [doc/] [html/] [manual/] [associative.html] - Blame information for rev 742

Details | Compare with Previous | View Log

Line No. Rev Author Line
1 742 jeremybenn
<?xml version="1.0" encoding="UTF-8" standalone="no"?>
2
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">
3
<html xmlns="http://www.w3.org/1999/xhtml"><head><title>Associative</title><meta name="generator" content="DocBook XSL-NS Stylesheets V1.76.1"/><meta name="keywords" content="&#10;      ISO C++&#10;    , &#10;      library&#10;    "/><meta name="keywords" content="&#10;      ISO C++&#10;    , &#10;      runtime&#10;    , &#10;      library&#10;    "/><link rel="home" href="../index.html" title="The GNU C++ Library"/><link rel="up" href="containers.html" title="Chapter 9.  Containers"/><link rel="prev" href="containers.html" title="Chapter 9.  Containers"/><link rel="next" href="containers_and_c.html" title="Interacting with C"/></head><body><div class="navheader"><table width="100%" summary="Navigation header"><tr><th colspan="3" align="center">Associative</th></tr><tr><td align="left"><a accesskey="p" href="containers.html">Prev</a> </td><th width="60%" align="center">Chapter 9. 
4
  Containers
5
 
6
</th><td align="right"> <a accesskey="n" href="containers_and_c.html">Next</a></td></tr></table><hr/></div><div class="section" title="Associative"><div class="titlepage"><div><div><h2 class="title"><a id="std.containers.associative"/>Associative</h2></div></div></div><div class="section" title="Insertion Hints"><div class="titlepage"><div><div><h3 class="title"><a id="containers.associative.insert_hints"/>Insertion Hints</h3></div></div></div><p>
7
     Section [23.1.2], Table 69, of the C++ standard lists this
8
     function for all of the associative containers (map, set, etc):
9
   </p><pre class="programlisting">
10
      a.insert(p,t);
11
   </pre><p>
12
     where 'p' is an iterator into the container 'a', and 't' is the
13
     item to insert.  The standard says that <span class="quote">“<span class="quote"><code class="code">t</code> is
14
     inserted as close as possible to the position just prior to
15
     <code class="code">p</code>.</span>”</span> (Library DR #233 addresses this topic,
16
     referring to <a class="link" href="http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2005/n1780.html">N1780</a>.
17
     Since version 4.2 GCC implements the resolution to DR 233, so
18
     that insertions happen as close as possible to the hint. For
19
     earlier releases the hint was only used as described below.
20
   </p><p>
21
     Here we'll describe how the hinting works in the libstdc++
22
     implementation, and what you need to do in order to take
23
     advantage of it.  (Insertions can change from logarithmic
24
     complexity to amortized constant time, if the hint is properly
25
     used.)  Also, since the current implementation is based on the
26
     SGI STL one, these points may hold true for other library
27
     implementations also, since the HP/SGI code is used in a lot of
28
     places.
29
   </p><p>
30
     In the following text, the phrases <span class="emphasis"><em>greater
31
     than</em></span> and <span class="emphasis"><em>less than</em></span> refer to the
32
     results of the strict weak ordering imposed on the container by
33
     its comparison object, which defaults to (basically)
34
     <span class="quote">“<span class="quote">&lt;</span>”</span>.  Using those phrases is semantically sloppy,
35
     but I didn't want to get bogged down in syntax.  I assume that if
36
     you are intelligent enough to use your own comparison objects,
37
     you are also intelligent enough to assign <span class="quote">“<span class="quote">greater</span>”</span>
38
     and <span class="quote">“<span class="quote">lesser</span>”</span> their new meanings in the next
39
     paragraph.  *grin*
40
   </p><p>
41
     If the <code class="code">hint</code> parameter ('p' above) is equivalent to:
42
   </p><div class="itemizedlist"><ul class="itemizedlist"><li class="listitem"><p>
43
          <code class="code">begin()</code>, then the item being inserted should
44
          have a key less than all the other keys in the container.
45
          The item will be inserted at the beginning of the container,
46
          becoming the new entry at <code class="code">begin()</code>.
47
      </p></li><li class="listitem"><p>
48
          <code class="code">end()</code>, then the item being inserted should have
49
          a key greater than all the other keys in the container.  The
50
          item will be inserted at the end of the container, becoming
51
          the new entry before <code class="code">end()</code>.
52
      </p></li><li class="listitem"><p>
53
          neither <code class="code">begin()</code> nor <code class="code">end()</code>, then:
54
          Let <code class="code">h</code> be the entry in the container pointed to
55
          by <code class="code">hint</code>, that is, <code class="code">h = *hint</code>.  Then
56
          the item being inserted should have a key less than that of
57
          <code class="code">h</code>, and greater than that of the item preceding
58
          <code class="code">h</code>.  The new item will be inserted between
59
          <code class="code">h</code> and <code class="code">h</code>'s predecessor.
60
          </p></li></ul></div><p>
61
     For <code class="code">multimap</code> and <code class="code">multiset</code>, the
62
     restrictions are slightly looser: <span class="quote">“<span class="quote">greater than</span>”</span>
63
     should be replaced by <span class="quote">“<span class="quote">not less than</span>”</span>and <span class="quote">“<span class="quote">less
64
     than</span>”</span> should be replaced by <span class="quote">“<span class="quote">not greater
65
     than.</span>”</span> (Why not replace greater with
66
     greater-than-or-equal-to?  You probably could in your head, but
67
     the mathematicians will tell you that it isn't the same thing.)
68
   </p><p>
69
     If the conditions are not met, then the hint is not used, and the
70
     insertion proceeds as if you had called <code class="code"> a.insert(t)
71
     </code> instead.  (<span class="emphasis"><em>Note </em></span> that GCC releases
72
     prior to 3.0.2 had a bug in the case with <code class="code">hint ==
73
     begin()</code> for the <code class="code">map</code> and <code class="code">set</code>
74
     classes.  You should not use a hint argument in those releases.)
75
   </p><p>
76
     This behavior goes well with other containers'
77
     <code class="code">insert()</code> functions which take an iterator: if used,
78
     the new item will be inserted before the iterator passed as an
79
     argument, same as the other containers.
80
   </p><p>
81
     <span class="emphasis"><em>Note </em></span> also that the hint in this
82
     implementation is a one-shot.  The older insertion-with-hint
83
     routines check the immediately surrounding entries to ensure that
84
     the new item would in fact belong there.  If the hint does not
85
     point to the correct place, then no further local searching is
86
     done; the search begins from scratch in logarithmic time.
87
   </p></div><div class="section" title="bitset"><div class="titlepage"><div><div><h3 class="title"><a id="containers.associative.bitset"/>bitset</h3></div></div></div><div class="section" title="Size Variable"><div class="titlepage"><div><div><h4 class="title"><a id="associative.bitset.size_variable"/>Size Variable</h4></div></div></div><p>
88
        No, you cannot write code of the form
89
      </p><pre class="programlisting">
90
      #include &lt;bitset&gt;
91
 
92
      void foo (size_t n)
93
      {
94
          std::bitset&lt;n&gt;   bits;
95
          ....
96
      }
97
   </pre><p>
98
     because <code class="code">n</code> must be known at compile time.  Your
99
     compiler is correct; it is not a bug.  That's the way templates
100
     work.  (Yes, it <span class="emphasis"><em>is</em></span> a feature.)
101
   </p><p>
102
     There are a couple of ways to handle this kind of thing.  Please
103
     consider all of them before passing judgement.  They include, in
104
     no chaptericular order:
105
   </p><div class="itemizedlist"><ul class="itemizedlist"><li class="listitem"><p>A very large N in <code class="code">bitset&lt;N&gt;</code>.</p></li><li class="listitem"><p>A container&lt;bool&gt;.</p></li><li class="listitem"><p>Extremely weird solutions.</p></li></ul></div><p>
106
     <span class="emphasis"><em>A very large N in
107
     <code class="code">bitset&lt;N&gt;</code>.  </em></span> It has been
108
     pointed out a few times in newsgroups that N bits only takes up
109
     (N/8) bytes on most systems, and division by a factor of eight is
110
     pretty impressive when speaking of memory.  Half a megabyte given
111
     over to a bitset (recall that there is zero space overhead for
112
     housekeeping info; it is known at compile time exactly how large
113
     the set is) will hold over four million bits.  If you're using
114
     those bits as status flags (e.g.,
115
     <span class="quote">“<span class="quote">changed</span>”</span>/<span class="quote">“<span class="quote">unchanged</span>”</span> flags), that's a
116
     <span class="emphasis"><em>lot</em></span> of state.
117
   </p><p>
118
     You can then keep track of the <span class="quote">“<span class="quote">maximum bit used</span>”</span>
119
     during some testing runs on representative data, make note of how
120
     many of those bits really need to be there, and then reduce N to
121
     a smaller number.  Leave some extra space, of course.  (If you
122
     plan to write code like the incorrect example above, where the
123
     bitset is a local variable, then you may have to talk your
124
     compiler into allowing that much stack space; there may be zero
125
     space overhead, but it's all allocated inside the object.)
126
   </p><p>
127
     <span class="emphasis"><em>A container&lt;bool&gt;.  </em></span> The
128
     Committee made provision for the space savings possible with that
129
     (N/8) usage previously mentioned, so that you don't have to do
130
     wasteful things like <code class="code">Container&lt;char&gt;</code> or
131
     <code class="code">Container&lt;short int&gt;</code>.  Specifically,
132
     <code class="code">vector&lt;bool&gt;</code> is required to be specialized for
133
     that space savings.
134
   </p><p>
135
     The problem is that <code class="code">vector&lt;bool&gt;</code> doesn't
136
     behave like a normal vector anymore.  There have been
137
     journal articles which discuss the problems (the ones by Herb
138
     Sutter in the May and July/August 1999 issues of C++ Report cover
139
     it well).  Future revisions of the ISO C++ Standard will change
140
     the requirement for <code class="code">vector&lt;bool&gt;</code>
141
     specialization.  In the meantime, <code class="code">deque&lt;bool&gt;</code>
142
     is recommended (although its behavior is sane, you probably will
143
     not get the space savings, but the allocation scheme is different
144
     than that of vector).
145
   </p><p>
146
     <span class="emphasis"><em>Extremely weird solutions.  </em></span> If
147
     you have access to the compiler and linker at runtime, you can do
148
     something insane, like figuring out just how many bits you need,
149
     then writing a temporary source code file.  That file contains an
150
     instantiation of <code class="code">bitset</code> for the required number of
151
     bits, inside some wrapper functions with unchanging signatures.
152
     Have your program then call the compiler on that file using
153
     Position Independent Code, then open the newly-created object
154
     file and load those wrapper functions.  You'll have an
155
     instantiation of <code class="code">bitset&lt;N&gt;</code> for the exact
156
     <code class="code">N</code> that you need at the time.  Don't forget to delete
157
     the temporary files.  (Yes, this <span class="emphasis"><em>can</em></span> be, and
158
     <span class="emphasis"><em>has been</em></span>, done.)
159
   </p><p>
160
     This would be the approach of either a visionary genius or a
161
     raving lunatic, depending on your programming and management
162
     style.  Probably the latter.
163
   </p><p>
164
     Which of the above techniques you use, if any, are up to you and
165
     your intended application.  Some time/space profiling is
166
     indicated if it really matters (don't just guess).  And, if you
167
     manage to do anything along the lines of the third category, the
168
     author would love to hear from you...
169
   </p><p>
170
     Also note that the implementation of bitset used in libstdc++ has
171
     <a class="link" href="ext_containers.html#manual.ext.containers.sgi" title="Backwards Compatibility">some extensions</a>.
172
   </p></div><div class="section" title="Type String"><div class="titlepage"><div><div><h4 class="title"><a id="associative.bitset.type_string"/>Type String</h4></div></div></div><p>
173
      </p><p>
174
     Bitmasks do not take char* nor const char* arguments in their
175
     constructors.  This is something of an accident, but you can read
176
     about the problem: follow the library's <span class="quote">“<span class="quote">Links</span>”</span> from
177
     the homepage, and from the C++ information <span class="quote">“<span class="quote">defect
178
     reflector</span>”</span> link, select the library issues list.  Issue
179
     number 116 describes the problem.
180
   </p><p>
181
     For now you can simply make a temporary string object using the
182
     constructor expression:
183
   </p><pre class="programlisting">
184
      std::bitset&lt;5&gt; b ( std::string(<span class="quote">“<span class="quote">10110</span>”</span>) );
185
   </pre><p>
186
     instead of
187
   </p><pre class="programlisting">
188
      std::bitset&lt;5&gt; b ( <span class="quote">“<span class="quote">10110</span>”</span> );    // invalid
189
    </pre></div></div></div><div class="navfooter"><hr/><table width="100%" summary="Navigation footer"><tr><td align="left"><a accesskey="p" href="containers.html">Prev</a> </td><td align="center"><a accesskey="u" href="containers.html">Up</a></td><td align="right"> <a accesskey="n" href="containers_and_c.html">Next</a></td></tr><tr><td align="left" valign="top">Chapter 9. 
190
  Containers
191
 
192
 </td><td align="center"><a accesskey="h" href="../index.html">Home</a></td><td align="right" valign="top"> Interacting with C</td></tr></table></div></body></html>

powered by: WebSVN 2.1.0

© copyright 1999-2024 OpenCores.org, equivalent to Oliscience, all rights reserved. OpenCores®, registered trademark.