libstdc++-v3/doc/html/manual/bk01pt03ch17s04.html

   1 <?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>Design</title><meta name="generator" content="DocBook XSL-NS Stylesheets V1.76.1"/><meta name="keywords" content="&#10;      C++&#10;    , &#10;      library&#10;    , &#10;      debug&#10;    "/><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="debug_mode.html" title="Chapter 17. Debug Mode"/><link rel="prev" href="bk01pt03ch17s03.html" title="Using"/><link rel="next" href="parallel_mode.html" title="Chapter 18. Parallel Mode"/></head><body><div class="navheader"><table width="100%" summary="Navigation header"><tr><th colspan="3" align="center">Design</th></tr><tr><td align="left"><a accesskey="p" href="bk01pt03ch17s03.html">Prev</a> </td><th width="60%" align="center">Chapter 17. Debug Mode</th><td align="right"> <a accesskey="n" href="parallel_mode.html">Next</a></td></tr></table><hr/></div><div class="section" title="Design"><div class="titlepage"><div><div><h2 class="title"><a id="manual.ext.debug_mode.design"/>Design</h2></div></div></div><p>
   4   </p><div class="section" title="Goals"><div class="titlepage"><div><div><h3 class="title"><a id="debug_mode.design.goals"/>Goals</h3></div></div></div><p>
   5     </p><p> The libstdc++ debug mode replaces unsafe (but efficient) standard
   6   containers and iterators with semantically equivalent safe standard
   7   containers and iterators to aid in debugging user programs. The
   8   following goals directed the design of the libstdc++ debug mode:</p><div class="itemizedlist"><ul class="itemizedlist"><li class="listitem"><p><span class="emphasis"><em>Correctness</em></span>: the libstdc++ debug mode must not change
   9     the semantics of the standard library for all cases specified in
  10     the ANSI/ISO C++ standard. The essence of this constraint is that
  11     any valid C++ program should behave in the same manner regardless
  12     of whether it is compiled with debug mode or release mode. In
  13     particular, entities that are defined in namespace std in release
  14     mode should remain defined in namespace std in debug mode, so that
  15     legal specializations of namespace std entities will remain
  16     valid. A program that is not valid C++ (e.g., invokes undefined
  17     behavior) is not required to behave similarly, although the debug
  18     mode will abort with a diagnostic when it detects undefined
  19     behavior.</p></li><li class="listitem"><p><span class="emphasis"><em>Performance</em></span>: the additional of the libstdc++ debug mode
  20     must not affect the performance of the library when it is compiled
  21     in release mode. Performance of the libstdc++ debug mode is
  22     secondary (and, in fact, will be worse than the release
  23     mode).</p></li><li class="listitem"><p><span class="emphasis"><em>Usability</em></span>: the libstdc++ debug mode should be easy to
  24     use. It should be easily incorporated into the user's development
  25     environment (e.g., by requiring only a single new compiler switch)
  26     and should produce reasonable diagnostics when it detects a
  27     problem with the user program. Usability also involves detection
  28     of errors when using the debug mode incorrectly, e.g., by linking
  29     a release-compiled object against a debug-compiled object if in
  30     fact the resulting program will not run correctly.</p></li><li class="listitem"><p><span class="emphasis"><em>Minimize recompilation</em></span>: While it is expected that
  31     users recompile at least part of their program to use debug
  32     mode, the amount of recompilation affects the
  33     detect-compile-debug turnaround time. This indirectly affects the
  34     usefulness of the debug mode, because debugging some applications
  35     may require rebuilding a large amount of code, which may not be
  36     feasible when the suspect code may be very localized. There are
  37     several levels of conformance to this requirement, each with its
  38     own usability and implementation characteristics. In general, the
  39     higher-numbered conformance levels are more usable (i.e., require
  40     less recompilation) but are more complicated to implement than
  41     the lower-numbered conformance levels.
  42       </p><div class="orderedlist"><ol class="orderedlist"><li class="listitem"><p><span class="emphasis"><em>Full recompilation</em></span>: The user must recompile his or
  43         her entire application and all C++ libraries it depends on,
  44         including the C++ standard library that ships with the
  45         compiler. This must be done even if only a small part of the
  46         program can use debugging features.</p></li><li class="listitem"><p><span class="emphasis"><em>Full user recompilation</em></span>: The user must recompile
  47         his or her entire application and all C++ libraries it depends
  48         on, but not the C++ standard library itself. This must be done
  49         even if only a small part of the program can use debugging
  50         features. This can be achieved given a full recompilation
  51         system by compiling two versions of the standard library when
  52         the compiler is installed and linking against the appropriate
  53         one, e.g., a multilibs approach.</p></li><li class="listitem"><p><span class="emphasis"><em>Partial recompilation</em></span>: The user must recompile the
  54         parts of his or her application and the C++ libraries it
  55         depends on that will use the debugging facilities
  56         directly. This means that any code that uses the debuggable
  57         standard containers would need to be recompiled, but code
  58         that does not use them (but may, for instance, use IOStreams)
  59         would not have to be recompiled.</p></li><li class="listitem"><p><span class="emphasis"><em>Per-use recompilation</em></span>: The user must recompile the
  60         parts of his or her application and the C++ libraries it
  61         depends on where debugging should occur, and any other code
  62         that interacts with those containers. This means that a set of
  63         translation units that accesses a particular standard
  64         container instance may either be compiled in release mode (no
  65         checking) or debug mode (full checking), but must all be
  66         compiled in the same way; a translation unit that does not see
  67         that standard container instance need not be recompiled. This
  68         also means that a translation unit <span class="emphasis"><em>A</em></span> that contains a
  69         particular instantiation
  70         (say, <code class="code">std::vector&lt;int&gt;</code>) compiled in release
  71         mode can be linked against a translation unit <span class="emphasis"><em>B</em></span> that
  72         contains the same instantiation compiled in debug mode (a
  73         feature not present with partial recompilation). While this
  74         behavior is technically a violation of the One Definition
  75         Rule, this ability tends to be very important in
  76         practice. The libstdc++ debug mode supports this level of
  77         recompilation. </p></li><li class="listitem"><p><span class="emphasis"><em>Per-unit recompilation</em></span>: The user must only
  78         recompile the translation units where checking should occur,
  79         regardless of where debuggable standard containers are
  80         used. This has also been dubbed "<code class="code">-g</code> mode",
  81         because the <code class="code">-g</code> compiler switch works in this way,
  82         emitting debugging information at a per--translation-unit
  83         granularity. We believe that this level of recompilation is in
  84         fact not possible if we intend to supply safe iterators, leave
  85         the program semantics unchanged, and not regress in
  86         performance under release mode because we cannot associate
  87         extra information with an iterator (to form a safe iterator)
  88         without either reserving that space in release mode
  89         (performance regression) or allocating extra memory associated
  90         with each iterator with <code class="code">new</code> (changes the program
  91         semantics).</p></li></ol></div><p>
  92     </p></li></ul></div></div><div class="section" title="Methods"><div class="titlepage"><div><div><h3 class="title"><a id="debug_mode.design.methods"/>Methods</h3></div></div></div><p>
  93     </p><p>This section provides an overall view of the design of the
  94   libstdc++ debug mode and details the relationship between design
  95   decisions and the stated design goals.</p><div class="section" title="The Wrapper Model"><div class="titlepage"><div><div><h4 class="title"><a id="debug_mode.design.methods.wrappers"/>The Wrapper Model</h4></div></div></div><p>The libstdc++ debug mode uses a wrapper model where the
  96   debugging versions of library components (e.g., iterators and
  97   containers) form a layer on top of the release versions of the
  98   library components. The debugging components first verify that the
  99   operation is correct (aborting with a diagnostic if an error is
 100   found) and will then forward to the underlying release-mode
 101   container that will perform the actual work. This design decision
 102   ensures that we cannot regress release-mode performance (because the
 103   release-mode containers are left untouched) and partially
 104   enables <a class="link" href="bk01pt03ch17s04.html#methods.coexistence.link" title="Link- and run-time coexistence of release- and debug-mode components">mixing debug and
 105   release code</a> at link time, although that will not be
 106   discussed at this time.</p><p>Two types of wrappers are used in the implementation of the debug
 107   mode: container wrappers and iterator wrappers. The two types of
 108   wrappers interact to maintain relationships between iterators and
 109   their associated containers, which are necessary to detect certain
 110   types of standard library usage errors such as dereferencing
 111   past-the-end iterators or inserting into a container using an
 112   iterator from a different container.</p><div class="section" title="Safe Iterators"><div class="titlepage"><div><div><h5 class="title"><a id="debug_mode.design.methods.safe_iter"/>Safe Iterators</h5></div></div></div><p>Iterator wrappers provide a debugging layer over any iterator that
 113   is attached to a particular container, and will manage the
 114   information detailing the iterator's state (singular,
 115   dereferenceable, etc.) and tracking the container to which the
 116   iterator is attached. Because iterators have a well-defined, common
 117   interface the iterator wrapper is implemented with the iterator
 118   adaptor class template <code class="code">__gnu_debug::_Safe_iterator</code>,
 119   which takes two template parameters:</p><div class="itemizedlist"><ul class="itemizedlist"><li class="listitem"><p><code class="code">Iterator</code>: The underlying iterator type, which must
 120     be either the <code class="code">iterator</code> or <code class="code">const_iterator</code>
 121     typedef from the sequence type this iterator can reference.</p></li><li class="listitem"><p><code class="code">Sequence</code>: The type of sequence that this iterator
 122   references. This sequence must be a safe sequence (discussed below)
 123   whose <code class="code">iterator</code> or <code class="code">const_iterator</code> typedef
 124   is the type of the safe iterator.</p></li></ul></div></div><div class="section" title="Safe Sequences (Containers)"><div class="titlepage"><div><div><h5 class="title"><a id="debug_mode.design.methods.safe_seq"/>Safe Sequences (Containers)</h5></div></div></div><p>Container wrappers provide a debugging layer over a particular
 125   container type. Because containers vary greatly in the member
 126   functions they support and the semantics of those member functions
 127   (especially in the area of iterator invalidation), container
 128   wrappers are tailored to the container they reference, e.g., the
 129   debugging version of <code class="code">std::list</code> duplicates the entire
 130   interface of <code class="code">std::list</code>, adding additional semantic
 131   checks and then forwarding operations to the
 132   real <code class="code">std::list</code> (a public base class of the debugging
 133   version) as appropriate. However, all safe containers inherit from
 134   the class template <code class="code">__gnu_debug::_Safe_sequence</code>,
 135   instantiated with the type of the safe container itself (an instance
 136   of the curiously recurring template pattern).</p><p>The iterators of a container wrapper will be
 137   <a class="link" href="bk01pt03ch17s04.html#debug_mode.design.methods.safe_iter" title="Safe Iterators">safe
 138   iterators</a> that reference sequences of this type and wrap the
 139   iterators provided by the release-mode base class. The debugging
 140   container will use only the safe iterators within its own interface
 141   (therefore requiring the user to use safe iterators, although this
 142   does not change correct user code) and will communicate with the
 143   release-mode base class with only the underlying, unsafe,
 144   release-mode iterators that the base class exports.</p><p> The debugging version of <code class="code">std::list</code> will have the
 145   following basic structure:</p><pre class="programlisting">
 146 template&lt;typename _Tp, typename _Allocator = allocator&lt;_Tp&gt;
 147   class debug-list :
 148     public release-list&lt;_Tp, _Allocator&gt;,
 149     public __gnu_debug::_Safe_sequence&lt;debug-list&lt;_Tp, _Allocator&gt; &gt;
 150   {
 151     typedef release-list&lt;_Tp, _Allocator&gt; _Base;
 152     typedef debug-list&lt;_Tp, _Allocator&gt;   _Self;
 153
 154   public:
 155     typedef __gnu_debug::_Safe_iterator&lt;typename _Base::iterator, _Self&gt;       iterator;
 156     typedef __gnu_debug::_Safe_iterator&lt;typename _Base::const_iterator, _Self&gt; const_iterator;
 157
 158     // duplicate std::list interface with debugging semantics
 159   };
 160 </pre></div></div><div class="section" title="Precondition Checking"><div class="titlepage"><div><div><h4 class="title"><a id="debug_mode.design.methods.precond"/>Precondition Checking</h4></div></div></div><p>The debug mode operates primarily by checking the preconditions of
 161   all standard library operations that it supports. Preconditions that
 162   are always checked (regardless of whether or not we are in debug
 163   mode) are checked via the <code class="code">__check_xxx</code> macros defined
 164   and documented in the source
 165   file <code class="code">include/debug/debug.h</code>. Preconditions that may or
 166   may not be checked, depending on the debug-mode
 167   macro <code class="code">_GLIBCXX_DEBUG</code>, are checked via
 168   the <code class="code">__requires_xxx</code> macros defined and documented in the
 169   same source file. Preconditions are validated using any additional
 170   information available at run-time, e.g., the containers that are
 171   associated with a particular iterator, the position of the iterator
 172   within those containers, the distance between two iterators that may
 173   form a valid range, etc. In the absence of suitable information,
 174   e.g., an input iterator that is not a safe iterator, these
 175   precondition checks will silently succeed.</p><p>The majority of precondition checks use the aforementioned macros,
 176   which have the secondary benefit of having prewritten debug
 177   messages that use information about the current status of the
 178   objects involved (e.g., whether an iterator is singular or what
 179   sequence it is attached to) along with some static information
 180   (e.g., the names of the function parameters corresponding to the
 181   objects involved). When not using these macros, the debug mode uses
 182   either the debug-mode assertion
 183   macro <code class="code">_GLIBCXX_DEBUG_ASSERT</code> , its pedantic
 184   cousin <code class="code">_GLIBCXX_DEBUG_PEDASSERT</code>, or the assertion
 185   check macro that supports more advance formulation of error
 186   messages, <code class="code">_GLIBCXX_DEBUG_VERIFY</code>. These macros are
 187   documented more thoroughly in the debug mode source code.</p></div><div class="section" title="Release- and debug-mode coexistence"><div class="titlepage"><div><div><h4 class="title"><a id="debug_mode.design.methods.coexistence"/>Release- and debug-mode coexistence</h4></div></div></div><p>The libstdc++ debug mode is the first debug mode we know of that
 188   is able to provide the "Per-use recompilation" (4) guarantee, that
 189   allows release-compiled and debug-compiled code to be linked and
 190   executed together without causing unpredictable behavior. This
 191   guarantee minimizes the recompilation that users are required to
 192   perform, shortening the detect-compile-debug bug hunting cycle
 193   and making the debug mode easier to incorporate into development
 194   environments by minimizing dependencies.</p><p>Achieving link- and run-time coexistence is not a trivial
 195   implementation task. To achieve this goal we required a small
 196   extension to the GNU C++ compiler (since incorporated into the C++11 language specification, described in the GCC Manual for the C++ language as
 197   <a class="link" href="http://gcc.gnu.org/onlinedocs/gcc/Namespace-Association.html#Namespace-Association">namespace
 198   association</a>), and a complex organization of debug- and
 199   release-modes. The end result is that we have achieved per-use
 200   recompilation but have had to give up some checking of the
 201   <code class="code">std::basic_string</code> class template (namely, safe
 202   iterators).
 203 </p><div class="section" title="Compile-time coexistence of release- and debug-mode components"><div class="titlepage"><div><div><h5 class="title"><a id="methods.coexistence.compile"/>Compile-time coexistence of release- and debug-mode components</h5></div></div></div><p>Both the release-mode components and the debug-mode
 204   components need to exist within a single translation unit so that
 205   the debug versions can wrap the release versions. However, only one
 206   of these components should be user-visible at any particular
 207   time with the standard name, e.g., <code class="code">std::list</code>. </p><p>In release mode, we define only the release-mode version of the
 208   component with its standard name and do not include the debugging
 209   component at all. The release mode version is defined within the
 210   namespace <code class="code">std</code>. Minus the namespace associations, this
 211   method leaves the behavior of release mode completely unchanged from
 212   its behavior prior to the introduction of the libstdc++ debug
 213   mode. Here's an example of what this ends up looking like, in
 214   C++.</p><pre class="programlisting">
 215 namespace std
 216 {
 217   template&lt;typename _Tp, typename _Alloc = allocator&lt;_Tp&gt; &gt;
 218     class list
 219     {
 220       // ...
 221      };
 222 } // namespace std
 223 </pre><p>In debug mode we include the release-mode container (which is now
 224 defined in the namespace <code class="code">__cxx1998</code>) and also the
 225 debug-mode container. The debug-mode container is defined within the
 226 namespace <code class="code">__debug</code>, which is associated with namespace
 227 <code class="code">std</code> via the C++11 namespace association language feature.  This
 228 method allows the debug and release versions of the same component to
 229 coexist at compile-time and link-time without causing an unreasonable
 230 maintenance burden, while minimizing confusion. Again, this boils down
 231 to C++ code as follows:</p><pre class="programlisting">
 232 namespace std
 233 {
 234   namespace __cxx1998
 235   {
 236     template&lt;typename _Tp, typename _Alloc = allocator&lt;_Tp&gt; &gt;
 237       class list
 238       {
 239         // ...
 240       };
 241   } // namespace __gnu_norm
 242
 243   namespace __debug
 244   {
 245     template&lt;typename _Tp, typename _Alloc = allocator&lt;_Tp&gt; &gt;
 246       class list
 247       : public __cxx1998::list&lt;_Tp, _Alloc&gt;,
 248         public __gnu_debug::_Safe_sequence&lt;list&lt;_Tp, _Alloc&gt; &gt;
 249       {
 250         // ...
 251       };
 252   } // namespace __cxx1998
 253
 254   // namespace __debug __attribute__ ((strong));
 255   inline namespace __debug { }
 256 }
 257 </pre></div><div class="section" title="Link- and run-time coexistence of release- and debug-mode components"><div class="titlepage"><div><div><h5 class="title"><a id="methods.coexistence.link"/>Link- and run-time coexistence of release- and
 258     debug-mode components</h5></div></div></div><p>Because each component has a distinct and separate release and
 259 debug implementation, there is no issue with link-time
 260 coexistence: the separate namespaces result in different mangled
 261 names, and thus unique linkage.</p><p>However, components that are defined and used within the C++
 262 standard library itself face additional constraints. For instance,
 263 some of the member functions of <code class="code"> std::moneypunct</code> return
 264 <code class="code">std::basic_string</code>. Normally, this is not a problem, but
 265 with a mixed mode standard library that could be using either
 266 debug-mode or release-mode <code class="code"> basic_string</code> objects, things
 267 get more complicated.  As the return value of a function is not
 268 encoded into the mangled name, there is no way to specify a
 269 release-mode or a debug-mode string. In practice, this results in
 270 runtime errors. A simplified example of this problem is as follows.
 271 </p><p> Take this translation unit, compiled in debug-mode: </p><pre class="programlisting">
 272 // -D_GLIBCXX_DEBUG
 273 #include &lt;string&gt;
 274
 275 std::string test02();
 276
 277 std::string test01()
 278 {
 279   return test02();
 280 }
 281
 282 int main()
 283 {
 284   test01();
 285   return 0;
 286 }
 287 </pre><p> ... and linked to this translation unit, compiled in release mode:</p><pre class="programlisting">
 288 #include &lt;string&gt;
 289
 290 std::string
 291 test02()
 292 {
 293   return std::string("toast");
 294 }
 295 </pre><p> For this reason we cannot easily provide safe iterators for
 296   the <code class="code">std::basic_string</code> class template, as it is present
 297   throughout the C++ standard library. For instance, locale facets
 298   define typedefs that include <code class="code">basic_string</code>: in a mixed
 299   debug/release program, should that typedef be based on the
 300   debug-mode <code class="code">basic_string</code> or the
 301   release-mode <code class="code">basic_string</code>? While the answer could be
 302   "both", and the difference hidden via renaming a la the
 303   debug/release containers, we must note two things about locale
 304   facets:</p><div class="orderedlist"><ol class="orderedlist"><li class="listitem"><p>They exist as shared state: one can create a facet in one
 305   translation unit and access the facet via the same type name in a
 306   different translation unit. This means that we cannot have two
 307   different versions of locale facets, because the types would not be
 308   the same across debug/release-mode translation unit barriers.</p></li><li class="listitem"><p>They have virtual functions returning strings: these functions
 309   mangle in the same way regardless of the mangling of their return
 310   types (see above), and their precise signatures can be relied upon
 311   by users because they may be overridden in derived classes.</p></li></ol></div><p>With the design of libstdc++ debug mode, we cannot effectively hide
 312   the differences between debug and release-mode strings from the
 313   user. Failure to hide the differences may result in unpredictable
 314   behavior, and for this reason we have opted to only
 315   perform <code class="code">basic_string</code> changes that do not require ABI
 316   changes. The effect on users is expected to be minimal, as there are
 317   simple alternatives (e.g., <code class="code">__gnu_debug::basic_string</code>),
 318   and the usability benefit we gain from the ability to mix debug- and
 319   release-compiled translation units is enormous.</p></div><div class="section" title="Alternatives for Coexistence"><div class="titlepage"><div><div><h5 class="title"><a id="methods.coexistence.alt"/>Alternatives for Coexistence</h5></div></div></div><p>The coexistence scheme above was chosen over many alternatives,
 320   including language-only solutions and solutions that also required
 321   extensions to the C++ front end. The following is a partial list of
 322   solutions, with justifications for our rejection of each.</p><div class="itemizedlist"><ul class="itemizedlist"><li class="listitem"><p><span class="emphasis"><em>Completely separate debug/release libraries</em></span>: This is by
 323   far the simplest implementation option, where we do not allow any
 324   coexistence of debug- and release-compiled translation units in a
 325   program. This solution has an extreme negative affect on usability,
 326   because it is quite likely that some libraries an application
 327   depends on cannot be recompiled easily. This would not meet
 328   our <span class="emphasis"><em>usability</em></span> or <span class="emphasis"><em>minimize recompilation</em></span> criteria
 329   well.</p></li><li class="listitem"><p><span class="emphasis"><em>Add a <code class="code">Debug</code> boolean template parameter</em></span>:
 330   Partial specialization could be used to select the debug
 331   implementation when <code class="code">Debug == true</code>, and the state
 332   of <code class="code">_GLIBCXX_DEBUG</code> could decide whether the
 333   default <code class="code">Debug</code> argument is <code class="code">true</code>
 334   or <code class="code">false</code>. This option would break conformance with the
 335   C++ standard in both debug <span class="emphasis"><em>and</em></span> release modes. This would
 336   not meet our <span class="emphasis"><em>correctness</em></span> criteria. </p></li><li class="listitem"><p><span class="emphasis"><em>Packaging a debug flag in the allocators</em></span>: We could
 337     reuse the <code class="code">Allocator</code> template parameter of containers
 338     by adding a sentinel wrapper <code class="code">debug&lt;&gt;</code> that
 339     signals the user's intention to use debugging, and pick up
 340     the <code class="code">debug&lt;&gt;</code> allocator wrapper in a partial
 341     specialization. However, this has two drawbacks: first, there is a
 342     conformance issue because the default allocator would not be the
 343     standard-specified <code class="code">std::allocator&lt;T&gt;</code>. Secondly
 344     (and more importantly), users that specify allocators instead of
 345     implicitly using the default allocator would not get debugging
 346     containers. Thus this solution fails the <span class="emphasis"><em>correctness</em></span>
 347     criteria.</p></li><li class="listitem"><p><span class="emphasis"><em>Define debug containers in another namespace, and employ
 348       a <code class="code">using</code> declaration (or directive)</em></span>: This is an
 349       enticing option, because it would eliminate the need for
 350       the <code class="code">link_name</code> extension by aliasing the
 351       templates. However, there is no true template aliasing mechanism
 352       in C++, because both <code class="code">using</code> directives and using
 353       declarations disallow specialization. This method fails
 354       the <span class="emphasis"><em>correctness</em></span> criteria.</p></li><li class="listitem"><p><span class="emphasis"><em> Use implementation-specific properties of anonymous
 355     namespaces. </em></span>
 356     See <a class="link" href="http://gcc.gnu.org/ml/libstdc++/2003-08/msg00004.html"> this post
 357     </a>
 358     This method fails the <span class="emphasis"><em>correctness</em></span> criteria.</p></li><li class="listitem"><p><span class="emphasis"><em>Extension: allow reopening on namespaces</em></span>: This would
 359     allow the debug mode to effectively alias the
 360     namespace <code class="code">std</code> to an internal namespace, such
 361     as <code class="code">__gnu_std_debug</code>, so that it is completely
 362     separate from the release-mode <code class="code">std</code> namespace. While
 363     this will solve some renaming problems and ensure that
 364     debug- and release-compiled code cannot be mixed unsafely, it ensures that
 365     debug- and release-compiled code cannot be mixed at all. For
 366     instance, the program would have two <code class="code">std::cout</code>
 367     objects! This solution would fails the <span class="emphasis"><em>minimize
 368     recompilation</em></span> requirement, because we would only be able to
 369     support option (1) or (2).</p></li><li class="listitem"><p><span class="emphasis"><em>Extension: use link name</em></span>: This option involves
 370     complicated re-naming between debug-mode and release-mode
 371     components at compile time, and then a g++ extension called <span class="emphasis"><em>
 372     link name </em></span> to recover the original names at link time. There
 373     are two drawbacks to this approach. One, it's very verbose,
 374     relying on macro renaming at compile time and several levels of
 375     include ordering. Two, ODR issues remained with container member
 376     functions taking no arguments in mixed-mode settings resulting in
 377     equivalent link names, <code class="code"> vector::push_back() </code> being
 378     one example.
 379     See <a class="link" href="http://gcc.gnu.org/ml/libstdc++/2003-08/msg00177.html">link
 380     name</a> </p></li></ul></div><p>Other options may exist for implementing the debug mode, many of
 381   which have probably been considered and others that may still be
 382   lurking. This list may be expanded over time to include other
 383   options that we could have implemented, but in all cases the full
 384   ramifications of the approach (as measured against the design goals
 385   for a libstdc++ debug mode) should be considered first. The DejaGNU
 386   testsuite includes some testcases that check for known problems with
 387   some solutions (e.g., the <code class="code">using</code> declaration solution
 388   that breaks user specialization), and additional testcases will be
 389   added as we are able to identify other typical problem cases. These
 390   test cases will serve as a benchmark by which we can compare debug
 391   mode implementations.</p></div></div></div><div class="section" title="Other Implementations"><div class="titlepage"><div><div><h3 class="title"><a id="debug_mode.design.other"/>Other Implementations</h3></div></div></div><p>
 392     </p><p> There are several existing implementations of debug modes for C++
 393   standard library implementations, although none of them directly
 394   supports debugging for programs using libstdc++. The existing
 395   implementations include:</p><div class="itemizedlist"><ul class="itemizedlist"><li class="listitem"><p><a class="link" href="http://www.mathcs.sjsu.edu/faculty/horstman/safestl.html">SafeSTL</a>:
 396   SafeSTL was the original debugging version of the Standard Template
 397   Library (STL), implemented by Cay S. Horstmann on top of the
 398   Hewlett-Packard STL. Though it inspired much work in this area, it
 399   has not been kept up-to-date for use with modern compilers or C++
 400   standard library implementations.</p></li><li class="listitem"><p><a class="link" href="http://www.stlport.org/">STLport</a>: STLport is a free
 401   implementation of the C++ standard library derived from the <a class="link" href="http://www.sgi.com/tech/stl/">SGI implementation</a>, and
 402   ported to many other platforms. It includes a debug mode that uses a
 403   wrapper model (that in some ways inspired the libstdc++ debug mode
 404   design), although at the time of this writing the debug mode is
 405   somewhat incomplete and meets only the "Full user recompilation" (2)
 406   recompilation guarantee by requiring the user to link against a
 407   different library in debug mode vs. release mode.</p></li><li class="listitem"><p>Metrowerks CodeWarrior: The C++ standard library
 408   that ships with Metrowerks CodeWarrior includes a debug mode. It is
 409   a full debug-mode implementation (including debugging for
 410   CodeWarrior extensions) and is easy to use, although it meets only
 411   the "Full recompilation" (1) recompilation
 412   guarantee.</p></li></ul></div></div></div><div class="navfooter"><hr/><table width="100%" summary="Navigation footer"><tr><td align="left"><a accesskey="p" href="bk01pt03ch17s03.html">Prev</a> </td><td align="center"><a accesskey="u" href="debug_mode.html">Up</a></td><td align="right"> <a accesskey="n" href="parallel_mode.html">Next</a></td></tr><tr><td align="left" valign="top">Using </td><td align="center"><a accesskey="h" href="../index.html">Home</a></td><td align="right" valign="top"> Chapter 18. Parallel Mode</td></tr></table></div></body></html>