blob: def011e7408f2caaf25f8faa4f419bb95df9290c [file] [log] [blame]
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<html>
<head><meta http-equiv="Content-Type" content="text/html;charset=iso-8859-1">
<title>Tables</title>
</head>
<body bgcolor="#ffffff">
<h1>Tables</h1>
<p>Most of the requirements on containers are presented in the ISO standard
in the form of tables. In order to avoid massive duplication of effort
while documenting all the classes, we follow the standard's lead and
present the base information here. Individual classes will only document
their departures from these tables (removed functions, additional functions,
changes, etc).
</p>
<p>We will not try to duplicate all of the surrounding text (footnotes,
explanations, etc.) from the standard, because that would also entail a
duplication of effort. Some of the surrounding text has been paraphrased
here for clarity. If you are uncertain about the meaning or interpretation
of these notes, consult a good textbook, and/or purchase your own copy of
the standard (it's cheap, see our FAQ).
</p>
<p>The table numbers are the same as those used in the standard. Tables can
be jumped to using their number, e.g., &quot;tables.html#67&quot;. Only
Tables 65 through 69 are presented. Some of the active Defect Reports
are also noted or incorporated.
</p>
<hr />
<a name="65"><p>
<table cellpadding="3" cellspacing="5" align="center" rules="rows" border="3"
cols="5" title="Table 65">
<caption><h2>Table 65 --- Container Requirements</h2></caption>
<tr><th colspan="5">
Anything calling itself a container must meet these minimum requirements.
</th></tr>
<tr>
<td><strong>expression</strong></td>
<td><strong>result type</strong></td>
<td><strong>operational semantics</strong></td>
<td><strong>notes, pre-/post-conditions, assertions</strong></td>
<td><strong>complexity</strong></td>
</tr>
<tr>
<td>X::value_type</td>
<td>T</td>
<td>&nbsp;</td>
<td>T is Assignable</td>
<td>compile time</td>
</tr>
<tr>
<td>X::reference</td>
<td>lvalue of T</td>
<td>&nbsp;</td>
<td>&nbsp;</td>
<td>compile time</td>
</tr>
<tr>
<td>X::const_reference</td>
<td>const lvalue of T</td>
<td>&nbsp;</td>
<td>&nbsp;</td>
<td>compile time</td>
</tr>
<tr>
<td>X::iterator</td>
<td>iterator type pointing to T</td>
<td>&nbsp;</td>
<td>Any iterator category except output iterator.
Convertible to X::const_iterator.</td>
<td>compile time</td>
</tr>
<tr>
<td>X::const_iterator</td>
<td>iterator type pointing to const T</td>
<td>&nbsp;</td>
<td>Any iterator category except output iterator.</td>
<td>compile time</td>
</tr>
<tr>
<td>X::difference_type</td>
<td>signed integral type</td>
<td>&nbsp;</td>
<td>identical to the difference type of X::iterator and X::const_iterator</td>
<td>compile time</td>
</tr>
<tr>
<td>X::size_type</td>
<td>unsigned integral type</td>
<td>&nbsp;</td>
<td>size_type can represent any non-negative value of difference_type</td>
<td>compile time</td>
</tr>
<tr>
<td>X u;</td>
<td>&nbsp;</td>
<td>&nbsp;</td>
<td>post: u.size() == 0</td>
<td>constant</td>
</tr>
<tr>
<td>X();</td>
<td>&nbsp;</td>
<td>&nbsp;</td>
<td>X().size == 0</td>
<td>constant</td>
</tr>
<tr>
<td>X(a);</td>
<td>&nbsp;</td>
<td>&nbsp;</td>
<td>a == X(a)</td>
<td>linear</td>
</tr>
<tr>
<td>X u(a);<br />X u = a;</td>
<td>&nbsp;</td>
<td>&nbsp;</td>
<td>post: u == a. Equivalent to: X u; u = a;</td>
<td>linear</td>
</tr>
<tr>
<td>(&amp;a)-&gt;~X();</td>
<td>void</td>
<td>&nbsp;</td>
<td>dtor is applied to every element of a; all the memory is deallocated</td>
<td>linear</td>
</tr>
<tr>
<td>a.begin()</td>
<td>iterator; const_iterator for constant a</td>
<td>&nbsp;</td>
<td>&nbsp;</td>
<td>constant</td>
</tr>
<tr>
<td>a.end()</td>
<td>iterator; const_iterator for constant a</td>
<td>&nbsp;</td>
<td>&nbsp;</td>
<td>constant</td>
</tr>
<tr>
<td>a == b</td>
<td>convertible to bool</td>
<td>&nbsp;</td>
<td>== is an equivalence relation. a.size()==b.size() &amp;&amp;
equal(a.begin(),a.end(),b.begin())</td>
<td>linear</td>
</tr>
<tr>
<td>a != b</td>
<td>convertible to bool</td>
<td>&nbsp;</td>
<td>equivalent to !(a==b)</td>
<td>linear</td>
</tr>
<tr>
<td>a.swap(b)</td>
<td>void</td>
<td>&nbsp;</td>
<td>swap(a,b)</td>
<td>may or may not have constant complexity</td>
</tr>
<tr>
<td>r = a</td>
<td>X&amp;</td>
<td>&nbsp;</td>
<td>r == a</td>
<td>linear</td>
</tr>
<!-- a fifth column, "operation semantics," magically appears in the table
at this point... wtf? -->
<tr>
<td>a.size()</td>
<td>size_type</td>
<td>a.end() - a.begin()</td>
<td>&nbsp;</td>
<td>may or may not have constant complexity</td>
</tr>
<tr>
<td>a.max_size()</td>
<td>size_type</td>
<td>size() of the largest possible container</td>
<td>&nbsp;</td>
<td>may or may not have constant complexity</td>
</tr>
<tr>
<td>a.empty()</td>
<td>convertible to bool</td>
<td>a.size() == 0</td>
<td>&nbsp;</td>
<td>constant</td>
</tr>
<tr>
<td>a &lt; b</td>
<td>convertible to bool</td>
<td>lexographical_compare( a.begin, a.end(), b.begin(), b.end())</td>
<td>pre: &lt; is defined for T and is a total ordering relation</td>
<td>linear</td>
</tr>
<tr>
<td>a &gt; b</td>
<td>convertible to bool</td>
<td>b &lt; a</td>
<td>&nbsp;</td>
<td>linear</td>
</tr>
<tr>
<td>a &lt;= b</td>
<td>convertible to bool</td>
<td>!(a &gt; b)</td>
<td>&nbsp;</td>
<td>linear</td>
</tr>
<tr>
<td>a &gt;= b</td>
<td>convertible to bool</td>
<td>!(a &lt; b)</td>
<td>&nbsp;</td>
<td>linear</td>
</tr>
</table title="Table 65"></p></a>
<a name="66"><p>
<table cellpadding="3" cellspacing="5" align="center" rules="rows" border="3"
cols="4" title="Table 66">
<caption><h2>Table 66 --- Reversible Container Requirements</h2></caption>
<tr><th colspan="4">
If a container's iterator is bidirectional or random-access, then the
container also meets these requirements.
Deque, list, vector, map, multimap, set, and multiset are such containers.
</th></tr>
<tr>
<td><strong>expression</strong></td>
<td><strong>result type</strong></td>
<td><strong>notes, pre-/post-conditions, assertions</strong></td>
<td><strong>complexity</strong></td>
</tr>
<tr>
<td>X::reverse_iterator</td>
<td>iterator type pointing to T</td>
<td>reverse_iterator&lt;iterator&gt;</td>
<td>compile time</td>
</tr>
<tr>
<td>X::const_reverse_iterator</td>
<td>iterator type pointing to const T</td>
<td>reverse_iterator&lt;const_iterator&gt;</td>
<td>compile time</td>
</tr>
<tr>
<td>a.rbegin()</td>
<td>reverse_iterator; const_reverse_iterator for constant a</td>
<td>reverse_iterator(end())</td>
<td>constant</td>
</tr>
<tr>
<td>a.rend()</td>
<td>reverse_iterator; const_reverse_iterator for constant a</td>
<td>reverse_iterator(begin())</td>
<td>constant</td>
</tr>
</table title="Table 66"></p></a>
<a name="67"><p>
<table cellpadding="3" cellspacing="5" align="center" rules="rows" border="3"
cols="3" title="Table 67">
<caption><h2>Table 67 --- Sequence Requirements</h2></caption>
<tr><th colspan="3">
These are in addition to the requirements of <a href="#65">containers</a>.
Deque, list, and vector are such containers.
</th></tr>
<tr>
<td><strong>expression</strong></td>
<td><strong>result type</strong></td>
<td><strong>notes, pre-/post-conditions, assertions</strong></td>
</tr>
<tr>
<td>X(n,t)<br />X a(n,t)</td>
<td>&nbsp;</td>
<td>constructs a sequence with n copies of t<br />post: size() == n</td>
</tr>
<tr>
<td>X(i,j)<br />X a(i,j)</td>
<td>&nbsp;</td>
<td>constructs a sequence equal to the range [i,j)<br />
post: size() == distance(i,j)</td>
</tr>
<tr>
<td>a.insert(p,t)</td>
<td>iterator (points to the inserted copy of t)</td>
<td>inserts a copy of t before p</td>
</tr>
<tr>
<td>a.insert(p,n,t)</td>
<td>void</td>
<td>inserts n copies of t before p</td>
</tr>
<tr>
<td>a.insert(p,i,j)</td>
<td>void</td>
<td>inserts copies of elements in [i,j) before p<br />
pre: i, j are not iterators into a</td>
</tr>
<tr>
<td>a.erase(q)</td>
<td>iterator (points to the element following q (prior to erasure))</td>
<td>erases the element pointed to by q</td>
</tr>
<tr>
<td>a.erase(q1,q1)</td>
<td>iterator (points to the element pointed to by q2 (prior to erasure))</td>
<td>erases the elements in the range [q1,q2)</td>
</tr>
<tr>
<td>a.clear()</td>
<td>void</td>
<td>erase(begin(),end())<br />post: size() == 0</td>
</tr>
</table title="Table 67"></p></a>
<a name="68"><p>
<table cellpadding="3" cellspacing="5" align="center" rules="rows" border="3"
cols="4" title="Table 68">
<caption><h2>Table 68 --- Optional Sequence Operations</h2></caption>
<tr><th colspan="4">
These operations are only included in containers when the operation can be
done in constant time.
</th></tr>
<tr>
<td><strong>expression</strong></td>
<td><strong>result type</strong></td>
<td><strong>operational semantics</strong></td>
<td><strong>container</strong></td>
</tr>
<tr>
<td>a.front()</td>
<td>reference; const_reference for constant a</td>
<td>*a.begin()</td>
<td>vector, list, deque</td>
</tr>
<tr>
<td>a.back()</td>
<td>reference; const_reference for constant a</td>
<td>*--a.end()</td>
<td>vector, list, deque</td>
</tr>
<tr>
<td>a.push_front(x)</td>
<td>void</td>
<td>a.insert(a.begin(),x)</td>
<td>list, deque</td>
</tr>
<tr>
<td>a.push_back(x)</td>
<td>void</td>
<td>a.insert(a.end(),x)</td>
<td>vector, list, deque</td>
</tr>
<tr>
<td>a.pop_front()</td>
<td>void</td>
<td>a.erase(a.begin())</td>
<td>list, deque</td>
</tr>
<tr>
<td>a.pop_back()</td>
<td>void</td>
<td>a.erase(--a.end())</td>
<td>vector, list, deque</td>
</tr>
<tr>
<td>a[n]</td>
<td>reference; const_reference for constant a</td>
<td>*(a.begin() + n)</td>
<td>vector, deque</td>
</tr>
<tr>
<td>a.at(n)</td>
<td>reference; const_reference for constant a</td>
<td>*(a.begin() + n)<br />throws out_of_range if n&gt;=a.size()</td>
<td>vector, deque</td>
</tr>
</table title="Table 68"></p></a>
<a name="69"><p>
<table cellpadding="3" cellspacing="5" align="center" rules="rows" border="3"
cols="4" title="Table 69">
<caption><h2>Table 69 --- Associative Container Requirements</h2></caption>
<tr><th colspan="4">
These are in addition to the requirements of <a href="#65">containers</a>.
Map, multimap, set, and multiset are such containers. An associative
container supports <em>unique keys</em> (and is written as
<code>a_uniq</code> instead of <code>a</code>) if it may contain at most
one element for each key. Otherwise it supports <em>equivalent keys</em>
(and is written <code>a_eq</code>). Examples of the former are set and map,
examples of the latter are multiset and multimap.
</th></tr>
<tr>
<td><strong>expression</strong></td>
<td><strong>result type</strong></td>
<td><strong>notes, pre-/post-conditions, assertions</strong></td>
<td><strong>complexity</strong></td>
</tr>
<tr>
<td>X::key_type</td>
<td>Key</td>
<td>Key is Assignable</td>
<td>compile time</td>
</tr>
<tr>
<td>X::key_compare</td>
<td>Compare</td>
<td>defaults to less&lt;key_type&gt;</td>
<td>compile time</td>
</tr>
<tr>
<td>X::value_compare</td>
<td>a binary predicate type</td>
<td>same as key_compare for set and multiset; an ordering relation on
pairs induced by the first component (Key) for map and multimap</td>
<td>compile time</td>
</tr>
<tr>
<td>X(c)<br />X a(c)</td>
<td>&nbsp;</td>
<td>constructs an empty container which uses c as a comparison object</td>
<td>constant</td>
</tr>
<tr>
<td>X()<br />X a</td>
<td>&nbsp;</td>
<td>constructs an empty container using Compare() as a comparison object</td>
<td>constant</td>
</tr>
<tr>
<td>X(i,j,c)<br />X a(i,j,c)</td>
<td>&nbsp;</td>
<td>constructs an empty container and inserts elements from the range [i,j)
into it; uses c as a comparison object</td>
<td>NlogN in general where N is distance(i,j); linear if [i,j) is
sorted with value_comp()</td>
</tr>
<tr>
<td>X(i,j)<br />X a(i,j)</td>
<td>&nbsp;</td>
<td>same as previous, but uses Compare() as a comparison object</td>
<td>same as previous</td>
</tr>
<tr>
<td>a.key_comp()</td>
<td>X::key_compare</td>
<td>returns the comparison object out of which a was constructed</td>
<td>constant</td>
</tr>
<tr>
<td>a.value_comp()</td>
<td>X::value_compare</td>
<td>returns an object constructed out of the comparison object</td>
<td>constant</td>
</tr>
<tr>
<td>a_uniq.insert(t)</td>
<td>pair&lt;iterator,bool&gt;</td>
<td>&quot;Inserts t if and only if there is no element in the container with
key equivalent to the key of t. The bool component of the returned pair
is true -iff- the insertion took place, and the iterator component of
the pair points to the element with key equivalent to the key of
t.&quot;</td> <!-- DR 316 -->
<td>logarithmic</td>
</tr>
<tr>
<td>a_eq.insert(t)</td>
<td>iterator</td>
<td>inserts t, returns the iterator pointing to the inserted element</td>
<td>logarithmic</td>
</tr>
<tr>
<td>a.insert(p,t)</td>
<td>iterator</td>
<td>possibly inserts t (depending on whether a_uniq or a_eq); returns iterator
pointing to the element with key equivalent to the key of t; iterator p
is a hint pointing to where the insert should start to search</td>
<td>logarithmic in general, amortized constant if t is inserted right
after p<br />
<strong>[but see DR 233 and <a href="
http://gcc.gnu.org/onlinedocs/libstdc++/23_containers/howto.html#4">our
specific notes</a>]</strong></td>
</tr>
<tr>
<td>a.insert(i,j)</td>
<td>void</td>
<td>pre: i, j are not iterators into a. possibly inserts each element from
the range [i,j) (depending on whether a_uniq or a_eq)</td>
<td>Nlog(size()+N) where N is distance(i,j) in general</td> <!-- DR 264 -->
</tr>
<tr>
<td>a.erase(k)</td>
<td>size_type</td>
<td>erases all elements with key equivalent to k; returns number of erased
elements</td>
<td>log(size()) + count(k)</td>
</tr>
<tr>
<td>a.erase(q)</td>
<td>void</td>
<td>erases the element pointed to by q</td>
<td>amortized constant</td>
</tr>
<tr>
<td>a.erase(q1,q2)</td>
<td>void</td>
<td>erases all the elements in the range [q1,q2)</td>
<td>log(size()) + distance(q1,q2)</td>
</tr>
<tr>
<td>a.clear()</td>
<td>void</td>
<td>erases everything; post: size() == 0</td>
<td>linear</td> <!-- DR 224 -->
</tr>
<tr>
<td>a.find(k)</td>
<td>iterator; const_iterator for constant a</td>
<td>returns iterator pointing to element with key equivalent to k, or
a.end() if no such element found</td>
<td>logarithmic</td>
</tr>
<tr>
<td>a.count(k)</td>
<td>size_type</td>
<td>returns number of elements with key equivalent to k</td>
<td>log(size()) + count(k)</td>
</tr>
<tr>
<td>a.lower_bound(k)</td>
<td>iterator; const_iterator for constant a</td>
<td>returns iterator pointing to the first element with key not less than k</td>
<td>logarithmic</td>
</tr>
<tr>
<td>a.upper_bound(k)</td>
<td>iterator; const_iterator for constant a</td>
<td>returns iterator pointing to the first element with key greater than k</td>
<td>logarithmic</td>
</tr>
<tr>
<td>a.equal_range(k)</td>
<td>pair&lt;iterator,iterator&gt;;
pair&lt;const_iterator, const_iterator&gt; for constant a</td>
<td>equivalent to make_pair(a.lower_bound(k), a.upper_bound(k))</td>
<td>logarithmic</td>
</tr>
</table title="Table 69"></p></a>
<hr />
<p class="smallertext"><em>
See <a href="mainpage.html">mainpage.html</a> for copying conditions.
See <a href="http://gcc.gnu.org/libstdc++/">the libstdc++ homepage</a>
for more information.
</em></p>
</body>
</html>