| 1 | <?xml version="1.0" encoding="ISO-8859-1"?>
|
|---|
| 2 | <!DOCTYPE html
|
|---|
| 3 | PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
|
|---|
| 4 | "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
|
|---|
| 5 |
|
|---|
| 6 | <html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
|
|---|
| 7 | <head>
|
|---|
| 8 | <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1" />
|
|---|
| 9 | <meta name="AUTHOR" content="pme@gcc.gnu.org (Phil Edwards)" />
|
|---|
| 10 | <meta name="KEYWORDS" content="HOWTO, libstdc++, GCC, g++, libg++, STL" />
|
|---|
| 11 | <meta name="DESCRIPTION" content="HOWTO for the libstdc++ chapter 23." />
|
|---|
| 12 | <meta name="GENERATOR" content="vi and eight fingers" />
|
|---|
| 13 | <title>libstdc++-v3 HOWTO: Chapter 23</title>
|
|---|
| 14 | <link rel="StyleSheet" href="../lib3styles.css" />
|
|---|
| 15 | </head>
|
|---|
| 16 | <body>
|
|---|
| 17 |
|
|---|
| 18 | <h1 class="centered"><a name="top">Chapter 23: Containers</a></h1>
|
|---|
| 19 |
|
|---|
| 20 | <p>Chapter 23 deals with container classes and what they offer.
|
|---|
| 21 | </p>
|
|---|
| 22 |
|
|---|
| 23 |
|
|---|
| 24 | <!-- ####################################################### -->
|
|---|
| 25 | <hr />
|
|---|
| 26 | <h1>Contents</h1>
|
|---|
| 27 | <ul>
|
|---|
| 28 | <li><a href="#1">Making code unaware of the container/array difference</a></li>
|
|---|
| 29 | <li><a href="#2">Variable-sized bitmasks</a></li>
|
|---|
| 30 | <li><a href="#3">Containers and multithreading</a></li>
|
|---|
| 31 | <li><a href="#4">"Hinting" during insertion</a></li>
|
|---|
| 32 | <li><a href="#5">Bitmasks and string arguments</a></li>
|
|---|
| 33 | <li><a href="#6"><code>std::list::size()</code> is O(n)!</a></li>
|
|---|
| 34 | <li><a href="#7">Space overhead management for vectors</a></li>
|
|---|
| 35 | </ul>
|
|---|
| 36 |
|
|---|
| 37 | <hr />
|
|---|
| 38 |
|
|---|
| 39 | <!-- ####################################################### -->
|
|---|
| 40 |
|
|---|
| 41 | <h2><a name="1">Making code unaware of the container/array difference</a></h2>
|
|---|
| 42 | <p>You're writing some code and can't decide whether to use builtin
|
|---|
| 43 | arrays or some kind of container. There are compelling reasons
|
|---|
| 44 | to use one of the container classes, but you're afraid that you'll
|
|---|
| 45 | eventually run into difficulties, change everything back to arrays,
|
|---|
| 46 | and then have to change all the code that uses those data types to
|
|---|
| 47 | keep up with the change.
|
|---|
| 48 | </p>
|
|---|
| 49 | <p>If your code makes use of the standard algorithms, this isn't as
|
|---|
| 50 | scary as it sounds. The algorithms don't know, nor care, about
|
|---|
| 51 | the kind of "container" on which they work, since the
|
|---|
| 52 | algorithms are only given endpoints to work with. For the container
|
|---|
| 53 | classes, these are iterators (usually <code>begin()</code> and
|
|---|
| 54 | <code>end()</code>, but not always). For builtin arrays, these are
|
|---|
| 55 | the address of the first element and the
|
|---|
| 56 | <a href="../24_iterators/howto.html#2">past-the-end</a> element.
|
|---|
| 57 | </p>
|
|---|
| 58 | <p>Some very simple wrapper functions can hide all of that from the
|
|---|
| 59 | rest of the code. For example, a pair of functions called
|
|---|
| 60 | <code>beginof</code> can be written, one that takes an array, another
|
|---|
| 61 | that takes a vector. The first returns a pointer to the first
|
|---|
| 62 | element, and the second returns the vector's <code>begin()</code>
|
|---|
| 63 | iterator.
|
|---|
| 64 | </p>
|
|---|
| 65 | <p>The functions should be made template functions, and should also
|
|---|
| 66 | be declared inline. As pointed out in the comments in the code
|
|---|
| 67 | below, this can lead to <code>beginof</code> being optimized out of
|
|---|
| 68 | existence, so you pay absolutely nothing in terms of increased
|
|---|
| 69 | code size or execution time.
|
|---|
| 70 | </p>
|
|---|
| 71 | <p>The result is that if all your algorithm calls look like
|
|---|
| 72 | </p>
|
|---|
| 73 | <pre>
|
|---|
| 74 | std::transform(beginof(foo), endof(foo), beginof(foo), SomeFunction);</pre>
|
|---|
| 75 | <p>then the type of foo can change from an array of ints to a vector
|
|---|
| 76 | of ints to a deque of ints and back again, without ever changing any
|
|---|
| 77 | client code.
|
|---|
| 78 | </p>
|
|---|
| 79 | <p>This author has a collection of such functions, called "*of"
|
|---|
| 80 | because they all extend the builtin "sizeof". It started
|
|---|
| 81 | with some Usenet discussions on a transparent way to find the length
|
|---|
| 82 | of an array. A simplified and much-reduced version for easier
|
|---|
| 83 | reading is <a href="wrappers_h.txt">given here</a>.
|
|---|
| 84 | </p>
|
|---|
| 85 | <p>Astute readers will notice two things at once: first, that the
|
|---|
| 86 | container class is still a <code>vector<T></code> instead of a
|
|---|
| 87 | more general <code>Container<T></code>. This would mean that
|
|---|
| 88 | three functions for <code>deque</code> would have to be added, another
|
|---|
| 89 | three for <code>list</code>, and so on. This is due to problems with
|
|---|
| 90 | getting template resolution correct; I find it easier just to
|
|---|
| 91 | give the extra three lines and avoid confusion.
|
|---|
| 92 | </p>
|
|---|
| 93 | <p>Second, the line
|
|---|
| 94 | </p>
|
|---|
| 95 | <pre>
|
|---|
| 96 | inline unsigned int lengthof (T (&)[sz]) { return sz; } </pre>
|
|---|
| 97 | <p>looks just weird! Hint: unused parameters can be left nameless.
|
|---|
| 98 | </p>
|
|---|
| 99 | <p>Return <a href="#top">to top of page</a> or
|
|---|
| 100 | <a href="../faq/index.html">to the FAQ</a>.
|
|---|
| 101 | </p>
|
|---|
| 102 |
|
|---|
| 103 | <hr />
|
|---|
| 104 | <h2><a name="2">Variable-sized bitmasks</a></h2>
|
|---|
| 105 | <p>No, you cannot write code of the form
|
|---|
| 106 | </p>
|
|---|
| 107 | <!-- Careful, the leading spaces in PRE show up directly. -->
|
|---|
| 108 | <pre>
|
|---|
| 109 | #include <bitset>
|
|---|
| 110 |
|
|---|
| 111 | void foo (size_t n)
|
|---|
| 112 | {
|
|---|
| 113 | std::bitset<n> bits;
|
|---|
| 114 | ....
|
|---|
| 115 | } </pre>
|
|---|
| 116 | <p>because <code>n</code> must be known at compile time. Your compiler is
|
|---|
| 117 | correct; it is not a bug. That's the way templates work. (Yes, it
|
|---|
| 118 | <em>is</em> a feature.)
|
|---|
| 119 | </p>
|
|---|
| 120 | <p>There are a couple of ways to handle this kind of thing. Please
|
|---|
| 121 | consider all of them before passing judgement. They include, in
|
|---|
| 122 | no particular order:
|
|---|
| 123 | </p>
|
|---|
| 124 | <ul>
|
|---|
| 125 | <li>A very large N in <code>bitset<N></code>.</li>
|
|---|
| 126 | <li>A container<bool>.</li>
|
|---|
| 127 | <li>Extremely weird solutions.</li>
|
|---|
| 128 | </ul>
|
|---|
| 129 | <p><strong>A very large N in
|
|---|
| 130 | <code>bitset<N></code>. </strong> It has
|
|---|
| 131 | been pointed out a few times in newsgroups that N bits only takes up
|
|---|
| 132 | (N/8) bytes on most systems, and division by a factor of eight is pretty
|
|---|
| 133 | impressive when speaking of memory. Half a megabyte given over to a
|
|---|
| 134 | bitset (recall that there is zero space overhead for housekeeping info;
|
|---|
| 135 | it is known at compile time exactly how large the set is) will hold over
|
|---|
| 136 | four million bits. If you're using those bits as status flags (e.g.,
|
|---|
| 137 | "changed"/"unchanged" flags), that's a <em>lot</em>
|
|---|
| 138 | of state.
|
|---|
| 139 | </p>
|
|---|
| 140 | <p>You can then keep track of the "maximum bit used" during some
|
|---|
| 141 | testing runs on representative data, make note of how many of those bits
|
|---|
| 142 | really need to be there, and then reduce N to a smaller number. Leave
|
|---|
| 143 | some extra space, of course. (If you plan to write code like the
|
|---|
| 144 | incorrect example above, where the bitset is a local variable, then you
|
|---|
| 145 | may have to talk your compiler into allowing that much stack space;
|
|---|
| 146 | there may be zero space overhead, but it's all allocated inside the
|
|---|
| 147 | object.)
|
|---|
| 148 | </p>
|
|---|
| 149 | <p><strong>A container<bool>. </strong> The Committee
|
|---|
| 150 | made provision
|
|---|
| 151 | for the space savings possible with that (N/8) usage previously mentioned,
|
|---|
| 152 | so that you don't have to do wasteful things like
|
|---|
| 153 | <code>Container<char></code> or
|
|---|
| 154 | <code>Container<short int></code>.
|
|---|
| 155 | Specifically, <code>vector<bool></code> is required to be
|
|---|
| 156 | specialized for that space savings.
|
|---|
| 157 | </p>
|
|---|
| 158 | <p>The problem is that <code>vector<bool></code> doesn't behave like a
|
|---|
| 159 | normal vector anymore. There have been recent journal articles which
|
|---|
| 160 | discuss the problems (the ones by Herb Sutter in the May and
|
|---|
| 161 | July/August 1999 issues of
|
|---|
| 162 | <u>C++ Report</u> cover it well). Future revisions of the ISO C++
|
|---|
| 163 | Standard will change the requirement for <code>vector<bool></code>
|
|---|
| 164 | specialization. In the meantime, <code>deque<bool></code> is
|
|---|
| 165 | recommended (although its behavior is sane, you probably will not get
|
|---|
| 166 | the space savings, but the allocation scheme is different than that
|
|---|
| 167 | of vector).
|
|---|
| 168 | </p>
|
|---|
| 169 | <p><strong>Extremely weird solutions. </strong> If you have
|
|---|
| 170 | access to
|
|---|
| 171 | the compiler and linker at runtime, you can do something insane, like
|
|---|
| 172 | figuring out just how many bits you need, then writing a temporary
|
|---|
| 173 | source code file. That file contains an instantiation of
|
|---|
| 174 | <code>bitset</code>
|
|---|
| 175 | for the required number of bits, inside some wrapper functions with
|
|---|
| 176 | unchanging signatures. Have your program then call the
|
|---|
| 177 | compiler on that file using Position Independent Code, then open the
|
|---|
| 178 | newly-created object file and load those wrapper functions. You'll have
|
|---|
| 179 | an instantiation of <code>bitset<N></code> for the exact
|
|---|
| 180 | <code>N</code>
|
|---|
| 181 | that you need at the time. Don't forget to delete the temporary files.
|
|---|
| 182 | (Yes, this <em>can</em> be, and <em>has been</em>, done.)
|
|---|
| 183 | </p>
|
|---|
| 184 | <!-- I wonder if this next paragraph will get me in trouble... -->
|
|---|
| 185 | <p>This would be the approach of either a visionary genius or a raving
|
|---|
| 186 | lunatic, depending on your programming and management style. Probably
|
|---|
| 187 | the latter.
|
|---|
| 188 | </p>
|
|---|
| 189 | <p>Which of the above techniques you use, if any, are up to you and your
|
|---|
| 190 | intended application. Some time/space profiling is indicated if it
|
|---|
| 191 | really matters (don't just guess). And, if you manage to do anything
|
|---|
| 192 | along the lines of the third category, the author would love to hear
|
|---|
| 193 | from you...
|
|---|
| 194 | </p>
|
|---|
| 195 | <p>Also note that the implementation of bitset used in libstdc++-v3 has
|
|---|
| 196 | <a href="../ext/sgiexts.html#ch23">some extensions</a>.
|
|---|
| 197 | </p>
|
|---|
| 198 | <p>Return <a href="#top">to top of page</a> or
|
|---|
| 199 | <a href="../faq/index.html">to the FAQ</a>.
|
|---|
| 200 | </p>
|
|---|
| 201 |
|
|---|
| 202 | <hr />
|
|---|
| 203 | <h2><a name="3">Containers and multithreading</a></h2>
|
|---|
| 204 | <p>This section discusses issues surrounding the design of
|
|---|
| 205 | multithreaded applications which use Standard C++ containers.
|
|---|
| 206 | All information in this section is current as of the gcc 3.0
|
|---|
| 207 | release and all later point releases. Although earlier gcc
|
|---|
| 208 | releases had a different approach to threading configuration and
|
|---|
| 209 | proper compilation, the basic code design rules presented here
|
|---|
| 210 | were similar. For information on all other aspects of
|
|---|
| 211 | multithreading as it relates to libstdc++, including details on
|
|---|
| 212 | the proper compilation of threaded code (and compatibility between
|
|---|
| 213 | threaded and non-threaded code), see Chapter 17.
|
|---|
| 214 | </p>
|
|---|
| 215 | <p>Two excellent pages to read when working with the Standard C++
|
|---|
| 216 | containers and threads are
|
|---|
| 217 | <a href="http://www.sgi.com/tech/stl/thread_safety.html">SGI's
|
|---|
| 218 | http://www.sgi.com/tech/stl/thread_safety.html</a> and
|
|---|
| 219 | <a href="http://www.sgi.com/tech/stl/Allocators.html">SGI's
|
|---|
| 220 | http://www.sgi.com/tech/stl/Allocators.html</a>.
|
|---|
| 221 | </p>
|
|---|
| 222 | <p><em>However, please ignore all discussions about the user-level
|
|---|
| 223 | configuration of the lock implementation inside the STL
|
|---|
| 224 | container-memory allocator on those pages. For the sake of this
|
|---|
| 225 | discussion, libstdc++-v3 configures the SGI STL implementation,
|
|---|
| 226 | not you. This is quite different from how gcc pre-3.0 worked.
|
|---|
| 227 | In particular, past advice was for people using g++ to
|
|---|
| 228 | explicitly define _PTHREADS or other macros or port-specific
|
|---|
| 229 | compilation options on the command line to get a thread-safe
|
|---|
| 230 | STL. This is no longer required for any port and should no
|
|---|
| 231 | longer be done unless you really know what you are doing and
|
|---|
| 232 | assume all responsibility.</em>
|
|---|
| 233 | </p>
|
|---|
| 234 | <p>Since the container implementation of libstdc++-v3 uses the SGI
|
|---|
| 235 | code, we use the same definition of thread safety as SGI when
|
|---|
| 236 | discussing design. A key point that beginners may miss is the
|
|---|
| 237 | fourth major paragraph of the first page mentioned above
|
|---|
| 238 | ("For most clients,"...), which points out that
|
|---|
| 239 | locking must nearly always be done outside the container, by
|
|---|
| 240 | client code (that'd be you, not us). There is a notable
|
|---|
| 241 | exceptions to this rule. Allocators called while a container or
|
|---|
| 242 | element is constructed uses an internal lock obtained and
|
|---|
| 243 | released solely within libstdc++-v3 code (in fact, this is the
|
|---|
| 244 | reason STL requires any knowledge of the thread configuration).
|
|---|
| 245 | </p>
|
|---|
| 246 | <p>For implementing a container which does its own locking, it is
|
|---|
| 247 | trivial to provide a wrapper class which obtains the lock (as
|
|---|
| 248 | SGI suggests), performs the container operation, and then
|
|---|
| 249 | releases the lock. This could be templatized <em>to a certain
|
|---|
| 250 | extent</em>, on the underlying container and/or a locking
|
|---|
| 251 | mechanism. Trying to provide a catch-all general template
|
|---|
| 252 | solution would probably be more trouble than it's worth.
|
|---|
| 253 | </p>
|
|---|
| 254 | <p>The STL implementation is currently configured to use the
|
|---|
| 255 | high-speed caching memory allocator. Some people like to
|
|---|
| 256 | test and/or normally run threaded programs with a different
|
|---|
| 257 | default. For all details about how to globally override this
|
|---|
| 258 | at application run-time see <a href="../ext/howto.html#3">here</a>.
|
|---|
| 259 | </p>
|
|---|
| 260 | <p>There is a better way (not standardized yet): It is possible to
|
|---|
| 261 | force the malloc-based allocator on a per-case-basis for some
|
|---|
| 262 | application code. The library team generally believes that this
|
|---|
| 263 | is a better way to tune an application for high-speed using this
|
|---|
| 264 | implementation of the STL. There is
|
|---|
| 265 | <a href="../ext/howto.html#3">more information on allocators here</a>.
|
|---|
| 266 | </p>
|
|---|
| 267 | <p>Return <a href="#top">to top of page</a> or
|
|---|
| 268 | <a href="../faq/index.html">to the FAQ</a>.
|
|---|
| 269 | </p>
|
|---|
| 270 |
|
|---|
| 271 | <hr />
|
|---|
| 272 | <h2><a name="4">"Hinting" during insertion</a></h2>
|
|---|
| 273 | <p>Section [23.1.2], Table 69, of the C++ standard lists this function
|
|---|
| 274 | for all of the associative containers (map, set, etc):
|
|---|
| 275 | </p>
|
|---|
| 276 | <pre>
|
|---|
| 277 | a.insert(p,t);</pre>
|
|---|
| 278 | <p>where 'p' is an iterator into the container 'a', and 't' is the item
|
|---|
| 279 | to insert. The standard says that "iterator p is a hint
|
|---|
| 280 | pointing to where the insert should start to search," but
|
|---|
| 281 | specifies nothing more. (LWG Issue #233, currently in review,
|
|---|
| 282 | addresses this topic, but I will ignore it here because it is not yet
|
|---|
| 283 | finalized.)
|
|---|
| 284 | </p>
|
|---|
| 285 | <p>Here we'll describe how the hinting works in the libstdc++-v3
|
|---|
| 286 | implementation, and what you need to do in order to take advantage of
|
|---|
| 287 | it. (Insertions can change from logarithmic complexity to amortized
|
|---|
| 288 | constant time, if the hint is properly used.) Also, since the current
|
|---|
| 289 | implementation is based on the SGI STL one, these points may hold true
|
|---|
| 290 | for other library implementations also, since the HP/SGI code is used
|
|---|
| 291 | in a lot of places.
|
|---|
| 292 | </p>
|
|---|
| 293 | <p>In the following text, the phrases <em>greater than</em> and <em>less
|
|---|
| 294 | than</em> refer to the results of the strict weak ordering imposed on
|
|---|
| 295 | the container by its comparison object, which defaults to (basically)
|
|---|
| 296 | "<". Using those phrases is semantically sloppy, but I
|
|---|
| 297 | didn't want to get bogged down in syntax. I assume that if you are
|
|---|
| 298 | intelligent enough to use your own comparison objects, you are also
|
|---|
| 299 | intelligent enough to assign "greater" and "lesser"
|
|---|
| 300 | their new meanings in the next paragraph. *grin*
|
|---|
| 301 | </p>
|
|---|
| 302 | <p>If the <code>hint</code> parameter ('p' above) is equivalent to:
|
|---|
| 303 | </p>
|
|---|
| 304 | <ul>
|
|---|
| 305 | <li><code>begin()</code>, then the item being inserted should have a key
|
|---|
| 306 | less than all the other keys in the container. The item will
|
|---|
| 307 | be inserted at the beginning of the container, becoming the new
|
|---|
| 308 | entry at <code>begin()</code>.
|
|---|
| 309 | </li>
|
|---|
| 310 | <li><code>end()</code>, then the item being inserted should have a key
|
|---|
| 311 | greater than all the other keys in the container. The item will
|
|---|
| 312 | be inserted at the end of the container, becoming the new entry
|
|---|
| 313 | at <code>end()</code>.
|
|---|
| 314 | </li>
|
|---|
| 315 | <li>neither <code>begin()</code> nor <code>end()</code>, then: Let <code>h</code>
|
|---|
| 316 | be the entry in the container pointed to by <code>hint</code>, that
|
|---|
| 317 | is, <code>h = *hint</code>. Then the item being inserted should have
|
|---|
| 318 | a key less than that of <code>h</code>, and greater than that of the
|
|---|
| 319 | item preceding <code>h</code>. The new item will be inserted
|
|---|
| 320 | between <code>h</code> and <code>h</code>'s predecessor.
|
|---|
| 321 | </li>
|
|---|
| 322 | </ul>
|
|---|
| 323 | <p>For <code>multimap</code> and <code>multiset</code>, the restrictions are
|
|---|
| 324 | slightly looser: "greater than" should be replaced by
|
|---|
| 325 | "not less than" and "less than" should be replaced
|
|---|
| 326 | by "not greater than." (Why not replace greater with
|
|---|
| 327 | greater-than-or-equal-to? You probably could in your head, but the
|
|---|
| 328 | mathematicians will tell you that it isn't the same thing.)
|
|---|
| 329 | </p>
|
|---|
| 330 | <p>If the conditions are not met, then the hint is not used, and the
|
|---|
| 331 | insertion proceeds as if you had called <code> a.insert(t) </code>
|
|---|
| 332 | instead. (<strong>Note </strong> that GCC releases prior to 3.0.2
|
|---|
| 333 | had a bug in the case with <code>hint == begin()</code> for the
|
|---|
| 334 | <code>map</code> and <code>set</code> classes. You should not use a hint
|
|---|
| 335 | argument in those releases.)
|
|---|
| 336 | </p>
|
|---|
| 337 | <p>This behavior goes well with other container's <code>insert()</code>
|
|---|
| 338 | functions which take an iterator: if used, the new item will be
|
|---|
| 339 | inserted before the iterator passed as an argument, same as the other
|
|---|
| 340 | containers. The exception
|
|---|
| 341 | (in a sense) is with a hint of <code>end()</code>: the new item will
|
|---|
| 342 | actually be inserted after <code>end()</code>, but it also becomes the
|
|---|
| 343 | new <code>end()</code>.
|
|---|
| 344 | </p>
|
|---|
| 345 | <p><strong>Note </strong> also that the hint in this implementation is a
|
|---|
| 346 | one-shot. The insertion-with-hint routines check the immediately
|
|---|
| 347 | surrounding entries to ensure that the new item would in fact belong
|
|---|
| 348 | there. If the hint does not point to the correct place, then no
|
|---|
| 349 | further local searching is done; the search begins from scratch in
|
|---|
| 350 | logarithmic time. (Further local searching would only increase the
|
|---|
| 351 | time required when the hint is too far off.)
|
|---|
| 352 | </p>
|
|---|
| 353 | <p>Return <a href="#top">to top of page</a> or
|
|---|
| 354 | <a href="../faq/index.html">to the FAQ</a>.
|
|---|
| 355 | </p>
|
|---|
| 356 |
|
|---|
| 357 | <hr />
|
|---|
| 358 | <h2><a name="5">Bitmasks and string arguments</a></h2>
|
|---|
| 359 | <p>Bitmasks do not take char* nor const char* arguments in their
|
|---|
| 360 | constructors. This is something of an accident, but you can read
|
|---|
| 361 | about the problem: follow the library's "Links" from the
|
|---|
| 362 | homepage, and from the C++ information "defect reflector"
|
|---|
| 363 | link, select the library issues list. Issue number 116 describes the
|
|---|
| 364 | problem.
|
|---|
| 365 | </p>
|
|---|
| 366 | <p>For now you can simply make a temporary string object using the
|
|---|
| 367 | constructor expression:
|
|---|
| 368 | </p>
|
|---|
| 369 | <pre>
|
|---|
| 370 | std::bitset<5> b ( std::string("10110") );
|
|---|
| 371 | </pre>
|
|---|
| 372 | instead of
|
|---|
| 373 | <pre>
|
|---|
| 374 | std::bitset<5> b ( "10110" ); // invalid
|
|---|
| 375 | </pre>
|
|---|
| 376 | <p>Return <a href="#top">to top of page</a> or
|
|---|
| 377 | <a href="../faq/index.html">to the FAQ</a>.
|
|---|
| 378 | </p>
|
|---|
| 379 |
|
|---|
| 380 | <hr />
|
|---|
| 381 | <h2><a name="6"><code>std::list::size()</code> is O(n)!</a></h2>
|
|---|
| 382 | <p>Yes it is, and that's okay. This is a decision that we preserved when
|
|---|
| 383 | we imported SGI's STL implementation. The following is quoted from
|
|---|
| 384 | <a href="http://www.sgi.com/tech/stl/FAQ.html">their FAQ</a>:
|
|---|
| 385 | </p>
|
|---|
| 386 | <blockquote>
|
|---|
| 387 | <p>The size() member function, for list and slist, takes time
|
|---|
| 388 | proportional to the number of elements in the list. This was a
|
|---|
| 389 | deliberate tradeoff. The only way to get a constant-time size() for
|
|---|
| 390 | linked lists would be to maintain an extra member variable containing
|
|---|
| 391 | the list's size. This would require taking extra time to update that
|
|---|
| 392 | variable (it would make splice() a linear time operation, for example),
|
|---|
| 393 | and it would also make the list larger. Many list algorithms don't
|
|---|
| 394 | require that extra word (algorithms that do require it might do better
|
|---|
| 395 | with vectors than with lists), and, when it is necessary to maintain
|
|---|
| 396 | an explicit size count, it's something that users can do themselves.
|
|---|
| 397 | </p>
|
|---|
| 398 | <p>This choice is permitted by the C++ standard. The standard says that
|
|---|
| 399 | size() "should" be constant time, and "should"
|
|---|
| 400 | does not mean the same thing as "shall". This is the
|
|---|
| 401 | officially recommended ISO wording for saying that an implementation
|
|---|
| 402 | is supposed to do something unless there is a good reason not to.
|
|---|
| 403 | </p>
|
|---|
| 404 | <p>One implication of linear time size(): you should never write
|
|---|
| 405 | </p>
|
|---|
| 406 | <pre>
|
|---|
| 407 | if (L.size() == 0)
|
|---|
| 408 | ...</pre>
|
|---|
| 409 | Instead, you should write
|
|---|
| 410 | <pre>
|
|---|
| 411 | if (L.empty())
|
|---|
| 412 | ...</pre>
|
|---|
| 413 | </blockquote>
|
|---|
| 414 | <p>Return <a href="#top">to top of page</a> or
|
|---|
| 415 | <a href="../faq/index.html">to the FAQ</a>.
|
|---|
| 416 | </p>
|
|---|
| 417 |
|
|---|
| 418 | <hr />
|
|---|
| 419 | <h2><a name="7">Space overhead management for vectors</a></h2>
|
|---|
| 420 | <p>In
|
|---|
| 421 | <a href="http://gcc.gnu.org/ml/libstdc++/2002-04/msg00105.html">this
|
|---|
| 422 | message to the list</a>, Daniel Kostecky announced work on an
|
|---|
| 423 | alternate form of <code>std::vector</code> that would support hints
|
|---|
| 424 | on the number of elements to be over-allocated. The design was also
|
|---|
| 425 | described, along with possible implementation choices.
|
|---|
| 426 | </p>
|
|---|
| 427 | <p>The first two alpha releases were announced
|
|---|
| 428 | <a href="http://gcc.gnu.org/ml/libstdc++/2002-07/msg00048.html">here</a>
|
|---|
| 429 | and
|
|---|
| 430 | <a href="http://gcc.gnu.org/ml/libstdc++/2002-07/msg00111.html">here</a>.
|
|---|
| 431 | The releases themselves are available at
|
|---|
| 432 | <a href="http://www.kotelna.sk/dk/sw/caphint/">
|
|---|
| 433 | http://www.kotelna.sk/dk/sw/caphint/</a>.
|
|---|
| 434 | </p>
|
|---|
| 435 | <p>Return <a href="#top">to top of page</a> or
|
|---|
| 436 | <a href="../faq/index.html">to the FAQ</a>.
|
|---|
| 437 | </p>
|
|---|
| 438 |
|
|---|
| 439 |
|
|---|
| 440 | <!-- ####################################################### -->
|
|---|
| 441 |
|
|---|
| 442 | <hr />
|
|---|
| 443 | <p class="fineprint"><em>
|
|---|
| 444 | See <a href="../17_intro/license.html">license.html</a> for copying conditions.
|
|---|
| 445 | Comments and suggestions are welcome, and may be sent to
|
|---|
| 446 | <a href="mailto:libstdc++@gcc.gnu.org">the libstdc++ mailing list</a>.
|
|---|
| 447 | </em></p>
|
|---|
| 448 |
|
|---|
| 449 |
|
|---|
| 450 | </body>
|
|---|
| 451 | </html>
|
|---|
