gcc-4_8-branch/libstdc++-v3/doc/html/manual/debug_mode_design.html

   1 <?xml version="1.0" encoding="UTF-8" standalone="no"?>
   2 <!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>Design</title><meta name="generator" content="DocBook XSL-NS Stylesheets V1.78.1" /><meta name="keywords" content="C++, library, debug" /><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="debug_mode.html" title="Chapter 17. Debug Mode" /><link rel="prev" href="debug_mode_using.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 width="20%" align="left"><a accesskey="p" href="debug_mode_using.html">Prev</a> </td><th width="60%" align="center">Chapter 17. Debug Mode</th><td width="20%" align="right"> <a accesskey="n" href="parallel_mode.html">Next</a></td></tr></table><hr /></div><div class="section"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a id="manual.ext.debug_mode.design"></a>Design</h2></div></div></div><p>
   3   </p><div class="section"><div class="titlepage"><div><div><h3 class="title"><a id="debug_mode.design.goals"></a>Goals</h3></div></div></div><p>
   4     </p><p> The libstdc++ debug mode replaces unsafe (but efficient) standard
   5   containers and iterators with semantically equivalent safe standard
   6   containers and iterators to aid in debugging user programs. The
   7   following goals directed the design of the libstdc++ debug mode:</p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p><span class="emphasis"><em>Correctness</em></span>: the libstdc++ debug mode must not change
   8     the semantics of the standard library for all cases specified in
   9     the ANSI/ISO C++ standard. The essence of this constraint is that
  10     any valid C++ program should behave in the same manner regardless
  11     of whether it is compiled with debug mode or release mode. In
  12     particular, entities that are defined in namespace std in release
  13     mode should remain defined in namespace std in debug mode, so that
  14     legal specializations of namespace std entities will remain
  15     valid. A program that is not valid C++ (e.g., invokes undefined
  16     behavior) is not required to behave similarly, although the debug
  17     mode will abort with a diagnostic when it detects undefined
  18     behavior.</p></li><li class="listitem"><p><span class="emphasis"><em>Performance</em></span>: the additional of the libstdc++ debug mode
  19     must not affect the performance of the library when it is compiled
  20     in release mode. Performance of the libstdc++ debug mode is
  21     secondary (and, in fact, will be worse than the release
  22     mode).</p></li><li class="listitem"><p><span class="emphasis"><em>Usability</em></span>: the libstdc++ debug mode should be easy to
  23     use. It should be easily incorporated into the user's development
  24     environment (e.g., by requiring only a single new compiler switch)
  25     and should produce reasonable diagnostics when it detects a
  26     problem with the user program. Usability also involves detection
  27     of errors when using the debug mode incorrectly, e.g., by linking
  28     a release-compiled object against a debug-compiled object if in
  29     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
  30     users recompile at least part of their program to use debug
  31     mode, the amount of recompilation affects the
  32     detect-compile-debug turnaround time. This indirectly affects the
  33     usefulness of the debug mode, because debugging some applications
  34     may require rebuilding a large amount of code, which may not be
  35     feasible when the suspect code may be very localized. There are
  36     several levels of conformance to this requirement, each with its
  37     own usability and implementation characteristics. In general, the
  38     higher-numbered conformance levels are more usable (i.e., require
  39     less recompilation) but are more complicated to implement than
  40     the lower-numbered conformance levels.
  41       </p><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><p><span class="emphasis"><em>Full recompilation</em></span>: The user must recompile his or
  42         her entire application and all C++ libraries it depends on,
  43         including the C++ standard library that ships with the
  44         compiler. This must be done even if only a small part of the
  45         program can use debugging features.</p></li><li class="listitem"><p><span class="emphasis"><em>Full user recompilation</em></span>: The user must recompile
  46         his or her entire application and all C++ libraries it depends
  47         on, but not the C++ standard library itself. This must be done
  48         even if only a small part of the program can use debugging
  49         features. This can be achieved given a full recompilation
  50         system by compiling two versions of the standard library when
  51         the compiler is installed and linking against the appropriate
  52         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
  53         parts of his or her application and the C++ libraries it
  54         depends on that will use the debugging facilities
  55         directly. This means that any code that uses the debuggable
  56         standard containers would need to be recompiled, but code
  57         that does not use them (but may, for instance, use IOStreams)
  58         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
  59         parts of his or her application and the C++ libraries it
  60         depends on where debugging should occur, and any other code
  61         that interacts with those containers. This means that a set of
  62         translation units that accesses a particular standard
  63         container instance may either be compiled in release mode (no
  64         checking) or debug mode (full checking), but must all be
  65         compiled in the same way; a translation unit that does not see
  66         that standard container instance need not be recompiled. This
  67         also means that a translation unit <span class="emphasis"><em>A</em></span> that contains a
  68         particular instantiation
  69         (say, <code class="code">std::vector&lt;int&gt;</code>) compiled in release
  70         mode can be linked against a translation unit <span class="emphasis"><em>B</em></span> that
  71         contains the same instantiation compiled in debug mode (a
  72         feature not present with partial recompilation). While this
  73         behavior is technically a violation of the One Definition
  74         Rule, this ability tends to be very important in
  75         practice. The libstdc++ debug mode supports this level of
  76         recompilation. </p></li><li class="listitem"><p><span class="emphasis"><em>Per-unit recompilation</em></span>: The user must only
  77         recompile the translation units where checking should occur,
  78         regardless of where debuggable standard containers are
  79         used. This has also been dubbed "<code class="code">-g</code> mode",
  80         because the <code class="code">-g</code> compiler switch works in this way,
  81         emitting debugging information at a per--translation-unit
  82         granularity. We believe that this level of recompilation is in
  83         fact not possible if we intend to supply safe iterators, leave
  84         the program semantics unchanged, and not regress in
  85         performance under release mode because we cannot associate
  86         extra information with an iterator (to form a safe iterator)
  87         without either reserving that space in release mode
  88         (performance regression) or allocating extra memory associated
  89         with each iterator with <code class="code">new</code> (changes the program
  90         semantics).</p></li></ol></div><p>
  91     </p></li></ul></div></div><div class="section"><div class="titlepage"><div><div><h3 class="title"><a id="debug_mode.design.methods"></a>Methods</h3></div></div></div><p>
  92     </p><p>This section provides an overall view of the design of the
  93   libstdc++ debug mode and details the relationship between design
  94   decisions and the stated design goals.</p><div class="section"><div class="titlepage"><div><div><h4 class="title"><a id="debug_mode.design.methods.wrappers"></a>The Wrapper Model</h4></div></div></div><p>The libstdc++ debug mode uses a wrapper model where the
  95   debugging versions of library components (e.g., iterators and
  96   containers) form a layer on top of the release versions of the
  97   library components. The debugging components first verify that the
  98   operation is correct (aborting with a diagnostic if an error is
  99   found) and will then forward to the underlying release-mode
 100   container that will perform the actual work. This design decision
 101   ensures that we cannot regress release-mode performance (because the
 102   release-mode containers are left untouched) and partially
 103   enables <a class="link" href="debug_mode_design.html#methods.coexistence.link" title="Link- and run-time coexistence of release- and debug-mode components">mixing debug and
 104   release code</a> at link time, although that will not be
 105   discussed at this time.</p><p>Two types of wrappers are used in the implementation of the debug
 106   mode: container wrappers and iterator wrappers. The two types of
 107   wrappers interact to maintain relationships between iterators and
 108   their associated containers, which are necessary to detect certain
 109   types of standard library usage errors such as dereferencing
 110   past-the-end iterators or inserting into a container using an
 111   iterator from a different container.</p><div class="section"><div class="titlepage"><div><div><h5 class="title"><a id="debug_mode.design.methods.safe_iter"></a>Safe Iterators</h5></div></div></div><p>Iterator wrappers provide a debugging layer over any iterator that
 112   is attached to a particular container, and will manage the
 113   information detailing the iterator's state (singular,
 114   dereferenceable, etc.) and tracking the container to which the
 115   iterator is attached. Because iterators have a well-defined, common
 116   interface the iterator wrapper is implemented with the iterator
 117   adaptor class template <code class="code">__gnu_debug::_Safe_iterator</code>,
 118   which takes two template parameters:</p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p><code class="code">Iterator</code>: The underlying iterator type, which must
 119     be either the <code class="code">iterator</code> or <code class="code">const_iterator</code>
 120     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
 121   references. This sequence must be a safe sequence (discussed below)
 122   whose <code class="code">iterator</code> or <code class="code">const_iterator</code> typedef
 123   is the type of the safe iterator.</p></li></ul></div></div><div class="section"><div class="titlepage"><div><div><h5 class="title"><a id="debug_mode.design.methods.safe_seq"></a>Safe Sequences (Containers)</h5></div></div></div><p>Container wrappers provide a debugging layer over a particular
 124   container type. Because containers vary greatly in the member
 125   functions they support and the semantics of those member functions
 126   (especially in the area of iterator invalidation), container
 127   wrappers are tailored to the container they reference, e.g., the
 128   debugging version of <code class="code">std::list</code> duplicates the entire
 129   interface of <code class="code">std::list</code>, adding additional semantic
 130   checks and then forwarding operations to the
 131   real <code class="code">std::list</code> (a public base class of the debugging
 132   version) as appropriate. However, all safe containers inherit from
 133   the class template <code class="code">__gnu_debug::_Safe_sequence</code>,
 134   instantiated with the type of the safe container itself (an instance
 135   of the curiously recurring template pattern).</p><p>The iterators of a container wrapper will be
 136   <a class="link" href="debug_mode_design.html#debug_mode.design.methods.safe_iter" title="Safe Iterators">safe
 137   iterators</a> that reference sequences of this type and wrap the
 138   iterators provided by the release-mode base class. The debugging
 139   container will use only the safe iterators within its own interface
 140   (therefore requiring the user to use safe iterators, although this
 141   does not change correct user code) and will communicate with the
 142   release-mode base class with only the underlying, unsafe,
 143   release-mode iterators that the base class exports.</p><p> The debugging version of <code class="code">std::list</code> will have the
 144   following basic structure:</p><pre class="programlisting">
 145 template&lt;typename _Tp, typename _Allocator = allocator&lt;_Tp&gt;
 146   class debug-list :
 147     public release-list&lt;_Tp, _Allocator&gt;,
 148     public __gnu_debug::_Safe_sequence&lt;debug-list&lt;_Tp, _Allocator&gt; &gt;
 149   {
 150     typedef release-list&lt;_Tp, _Allocator&gt; _Base;
 151     typedef debug-list&lt;_Tp, _Allocator&gt;   _Self;
 152
 153   public:
 154     typedef __gnu_debug::_Safe_iterator&lt;typename _Base::iterator, _Self&gt;       iterator;
 155     typedef __gnu_debug::_Safe_iterator&lt;typename _Base::const_iterator, _Self&gt; const_iterator;
 156
 157     // duplicate std::list interface with debugging semantics
 158   };
 159 </pre></div></div><div class="section"><div class="titlepage"><div><div><h4 class="title"><a id="debug_mode.design.methods.precond"></a>Precondition Checking</h4></div></div></div><p>The debug mode operates primarily by checking the preconditions of
 160   all standard library operations that it supports. Preconditions that
 161   are always checked (regardless of whether or not we are in debug
 162   mode) are checked via the <code class="code">__check_xxx</code> macros defined
 163   and documented in the source
 164   file <code class="code">include/debug/debug.h</code>. Preconditions that may or
 165   may not be checked, depending on the debug-mode
 166   macro <code class="code">_GLIBCXX_DEBUG</code>, are checked via
 167   the <code class="code">__requires_xxx</code> macros defined and documented in the
 168   same source file. Preconditions are validated using any additional
 169   information available at run-time, e.g., the containers that are
 170   associated with a particular iterator, the position of the iterator
 171   within those containers, the distance between two iterators that may
 172   form a valid range, etc. In the absence of suitable information,
 173   e.g., an input iterator that is not a safe iterator, these
 174   precondition checks will silently succeed.</p><p>The majority of precondition checks use the aforementioned macros,
 175   which have the secondary benefit of having prewritten debug
 176   messages that use information about the current status of the
 177   objects involved (e.g., whether an iterator is singular or what
 178   sequence it is attached to) along with some static information
 179   (e.g., the names of the function parameters corresponding to the
 180   objects involved). When not using these macros, the debug mode uses
 181   either the debug-mode assertion
 182   macro <code class="code">_GLIBCXX_DEBUG_ASSERT</code> , its pedantic
 183   cousin <code class="code">_GLIBCXX_DEBUG_PEDASSERT</code>, or the assertion
 184   check macro that supports more advance formulation of error
 185   messages, <code class="code">_GLIBCXX_DEBUG_VERIFY</code>. These macros are
 186   documented more thoroughly in the debug mode source code.</p></div><div class="section"><div class="titlepage"><div><div><h4 class="title"><a id="debug_mode.design.methods.coexistence"></a>Release- and debug-mode coexistence</h4></div></div></div><p>The libstdc++ debug mode is the first debug mode we know of that
 187   is able to provide the "Per-use recompilation" (4) guarantee, that
 188   allows release-compiled and debug-compiled code to be linked and
 189   executed together without causing unpredictable behavior. This
 190   guarantee minimizes the recompilation that users are required to
 191   perform, shortening the detect-compile-debug bug hunting cycle
 192   and making the debug mode easier to incorporate into development
 193   environments by minimizing dependencies.</p><p>Achieving link- and run-time coexistence is not a trivial
 194   implementation task. To achieve this goal we required a small
 195   extension to the GNU C++ compiler (since incorporated into the C++11 language specification, described in the GCC Manual for the C++ language as
 196   <a class="link" href="http://gcc.gnu.org/onlinedocs/gcc/Namespace-Association.html#Namespace-Association" target="_top">namespace
 197   association</a>), and a complex organization of debug- and
 198   release-modes. The end result is that we have achieved per-use
 199   recompilation but have had to give up some checking of the
 200   <code class="code">std::basic_string</code> class template (namely, safe
 201   iterators).
 202 </p><div class="section"><div class="titlepage"><div><div><h5 class="title"><a id="methods.coexistence.compile"></a>Compile-time coexistence of release- and debug-mode components</h5></div></div></div><p>Both the release-mode components and the debug-mode
 203   components need to exist within a single translation unit so that
 204   the debug versions can wrap the release versions. However, only one
 205   of these components should be user-visible at any particular
 206   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
 207   component with its standard name and do not include the debugging
 208   component at all. The release mode version is defined within the
 209   namespace <code class="code">std</code>. Minus the namespace associations, this
 210   method leaves the behavior of release mode completely unchanged from
 211   its behavior prior to the introduction of the libstdc++ debug
 212   mode. Here's an example of what this ends up looking like, in
 213   C++.</p><pre class="programlisting">
 214 namespace std
 215 {
 216   template&lt;typename _Tp, typename _Alloc = allocator&lt;_Tp&gt; &gt;
 217     class list
 218     {
 219       // ...
 220      };
 221 } // namespace std
 222 </pre><p>In debug mode we include the release-mode container (which is now
 223 defined in the namespace <code class="code">__cxx1998</code>) and also the
 224 debug-mode container. The debug-mode container is defined within the
 225 namespace <code class="code">__debug</code>, which is associated with namespace
 226 <code class="code">std</code> via the C++11 namespace association language feature.  This
 227 method allows the debug and release versions of the same component to
 228 coexist at compile-time and link-time without causing an unreasonable
 229 maintenance burden, while minimizing confusion. Again, this boils down
 230 to C++ code as follows:</p><pre class="programlisting">
 231 namespace std
 232 {
 233   namespace __cxx1998
 234   {
 235     template&lt;typename _Tp, typename _Alloc = allocator&lt;_Tp&gt; &gt;
 236       class list
 237       {
 238         // ...
 239       };
 240   } // namespace __gnu_norm
 241
 242   namespace __debug
 243   {
 244     template&lt;typename _Tp, typename _Alloc = allocator&lt;_Tp&gt; &gt;
 245       class list
 246       : public __cxx1998::list&lt;_Tp, _Alloc&gt;,
 247         public __gnu_debug::_Safe_sequence&lt;list&lt;_Tp, _Alloc&gt; &gt;
 248       {
 249         // ...
 250       };
 251   } // namespace __cxx1998
 252
 253   // namespace __debug __attribute__ ((strong));
 254   inline namespace __debug { }
 255 }
 256 </pre></div><div class="section"><div class="titlepage"><div><div><h5 class="title"><a id="methods.coexistence.link"></a>Link- and run-time coexistence of release- and
 257     debug-mode components</h5></div></div></div><p>Because each component has a distinct and separate release and
 258 debug implementation, there is no issue with link-time
 259 coexistence: the separate namespaces result in different mangled
 260 names, and thus unique linkage.</p><p>However, components that are defined and used within the C++
 261 standard library itself face additional constraints. For instance,
 262 some of the member functions of <code class="code"> std::moneypunct</code> return
 263 <code class="code">std::basic_string</code>. Normally, this is not a problem, but
 264 with a mixed mode standard library that could be using either
 265 debug-mode or release-mode <code class="code"> basic_string</code> objects, things
 266 get more complicated.  As the return value of a function is not
 267 encoded into the mangled name, there is no way to specify a
 268 release-mode or a debug-mode string. In practice, this results in
 269 runtime errors. A simplified example of this problem is as follows.
 270 </p><p> Take this translation unit, compiled in debug-mode: </p><pre class="programlisting">
 271 // -D_GLIBCXX_DEBUG
 272 #include &lt;string&gt;
 273
 274 std::string test02();
 275
 276 std::string test01()
 277 {
 278   return test02();
 279 }
 280
 281 int main()
 282 {
 283   test01();
 284   return 0;
 285 }
 286 </pre><p> ... and linked to this translation unit, compiled in release mode:</p><pre class="programlisting">
 287 #include &lt;string&gt;
 288
 289 std::string
 290 test02()
 291 {
 292   return std::string("toast");
 293 }
 294 </pre><p> For this reason we cannot easily provide safe iterators for
 295   the <code class="code">std::basic_string</code> class template, as it is present
 296   throughout the C++ standard library. For instance, locale facets
 297   define typedefs that include <code class="code">basic_string</code>: in a mixed
 298   debug/release program, should that typedef be based on the
 299   debug-mode <code class="code">basic_string</code> or the
 300   release-mode <code class="code">basic_string</code>? While the answer could be
 301   "both", and the difference hidden via renaming a la the
 302   debug/release containers, we must note two things about locale
 303   facets:</p><div class="orderedlist"><ol class="orderedlist" type="1"><li class="listitem"><p>They exist as shared state: one can create a facet in one
 304   translation unit and access the facet via the same type name in a
 305   different translation unit. This means that we cannot have two
 306   different versions of locale facets, because the types would not be
 307   the same across debug/release-mode translation unit barriers.</p></li><li class="listitem"><p>They have virtual functions returning strings: these functions
 308   mangle in the same way regardless of the mangling of their return
 309   types (see above), and their precise signatures can be relied upon
 310   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
 311   the differences between debug and release-mode strings from the
 312   user. Failure to hide the differences may result in unpredictable
 313   behavior, and for this reason we have opted to only
 314   perform <code class="code">basic_string</code> changes that do not require ABI
 315   changes. The effect on users is expected to be minimal, as there are
 316   simple alternatives (e.g., <code class="code">__gnu_debug::basic_string</code>),
 317   and the usability benefit we gain from the ability to mix debug- and
 318   release-compiled translation units is enormous.</p></div><div class="section"><div class="titlepage"><div><div><h5 class="title"><a id="methods.coexistence.alt"></a>Alternatives for Coexistence</h5></div></div></div><p>The coexistence scheme above was chosen over many alternatives,
 319   including language-only solutions and solutions that also required
 320   extensions to the C++ front end. The following is a partial list of
 321   solutions, with justifications for our rejection of each.</p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p><span class="emphasis"><em>Completely separate debug/release libraries</em></span>: This is by
 322   far the simplest implementation option, where we do not allow any
 323   coexistence of debug- and release-compiled translation units in a
 324   program. This solution has an extreme negative affect on usability,
 325   because it is quite likely that some libraries an application
 326   depends on cannot be recompiled easily. This would not meet
 327   our <span class="emphasis"><em>usability</em></span> or <span class="emphasis"><em>minimize recompilation</em></span> criteria
 328   well.</p></li><li class="listitem"><p><span class="emphasis"><em>Add a <code class="code">Debug</code> boolean template parameter</em></span>:
 329   Partial specialization could be used to select the debug
 330   implementation when <code class="code">Debug == true</code>, and the state
 331   of <code class="code">_GLIBCXX_DEBUG</code> could decide whether the
 332   default <code class="code">Debug</code> argument is <code class="code">true</code>
 333   or <code class="code">false</code>. This option would break conformance with the
 334   C++ standard in both debug <span class="emphasis"><em>and</em></span> release modes. This would
 335   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
 336     reuse the <code class="code">Allocator</code> template parameter of containers
 337     by adding a sentinel wrapper <code class="code">debug&lt;&gt;</code> that
 338     signals the user's intention to use debugging, and pick up
 339     the <code class="code">debug&lt;&gt;</code> allocator wrapper in a partial
 340     specialization. However, this has two drawbacks: first, there is a
 341     conformance issue because the default allocator would not be the
 342     standard-specified <code class="code">std::allocator&lt;T&gt;</code>. Secondly
 343     (and more importantly), users that specify allocators instead of
 344     implicitly using the default allocator would not get debugging
 345     containers. Thus this solution fails the <span class="emphasis"><em>correctness</em></span>
 346     criteria.</p></li><li class="listitem"><p><span class="emphasis"><em>Define debug containers in another namespace, and employ
 347       a <code class="code">using</code> declaration (or directive)</em></span>: This is an
 348       enticing option, because it would eliminate the need for
 349       the <code class="code">link_name</code> extension by aliasing the
 350       templates. However, there is no true template aliasing mechanism
 351       in C++, because both <code class="code">using</code> directives and using
 352       declarations disallow specialization. This method fails
 353       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
 354     namespaces. </em></span>
 355     See <a class="link" href="http://gcc.gnu.org/ml/libstdc++/2003-08/msg00004.html" target="_top"> this post
 356     </a>
 357     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
 358     allow the debug mode to effectively alias the
 359     namespace <code class="code">std</code> to an internal namespace, such
 360     as <code class="code">__gnu_std_debug</code>, so that it is completely
 361     separate from the release-mode <code class="code">std</code> namespace. While
 362     this will solve some renaming problems and ensure that
 363     debug- and release-compiled code cannot be mixed unsafely, it ensures that
 364     debug- and release-compiled code cannot be mixed at all. For
 365     instance, the program would have two <code class="code">std::cout</code>
 366     objects! This solution would fails the <span class="emphasis"><em>minimize
 367     recompilation</em></span> requirement, because we would only be able to
 368     support option (1) or (2).</p></li><li class="listitem"><p><span class="emphasis"><em>Extension: use link name</em></span>: This option involves
 369     complicated re-naming between debug-mode and release-mode
 370     components at compile time, and then a g++ extension called <span class="emphasis"><em>
 371     link name </em></span> to recover the original names at link time. There
 372     are two drawbacks to this approach. One, it's very verbose,
 373     relying on macro renaming at compile time and several levels of
 374     include ordering. Two, ODR issues remained with container member
 375     functions taking no arguments in mixed-mode settings resulting in
 376     equivalent link names, <code class="code"> vector::push_back() </code> being
 377     one example.
 378     See <a class="link" href="http://gcc.gnu.org/ml/libstdc++/2003-08/msg00177.html" target="_top">link
 379     name</a> </p></li></ul></div><p>Other options may exist for implementing the debug mode, many of
 380   which have probably been considered and others that may still be
 381   lurking. This list may be expanded over time to include other
 382   options that we could have implemented, but in all cases the full
 383   ramifications of the approach (as measured against the design goals
 384   for a libstdc++ debug mode) should be considered first. The DejaGNU
 385   testsuite includes some testcases that check for known problems with
 386   some solutions (e.g., the <code class="code">using</code> declaration solution
 387   that breaks user specialization), and additional testcases will be
 388   added as we are able to identify other typical problem cases. These
 389   test cases will serve as a benchmark by which we can compare debug
 390   mode implementations.</p></div></div></div><div class="section"><div class="titlepage"><div><div><h3 class="title"><a id="debug_mode.design.other"></a>Other Implementations</h3></div></div></div><p>
 391     </p><p> There are several existing implementations of debug modes for C++
 392   standard library implementations, although none of them directly
 393   supports debugging for programs using libstdc++. The existing
 394   implementations include:</p><div class="itemizedlist"><ul class="itemizedlist" style="list-style-type: disc; "><li class="listitem"><p><a class="link" href="http://www.cs.sjsu.edu/faculty/horstman/safestl.html" target="_top">SafeSTL</a>:
 395   SafeSTL was the original debugging version of the Standard Template
 396   Library (STL), implemented by Cay S. Horstmann on top of the
 397   Hewlett-Packard STL. Though it inspired much work in this area, it
 398   has not been kept up-to-date for use with modern compilers or C++
 399   standard library implementations.</p></li><li class="listitem"><p><a class="link" href="http://www.stlport.org/" target="_top">STLport</a>: STLport is a free
 400   implementation of the C++ standard library derived from the <a class="link" href="http://www.sgi.com/tech/stl/" target="_top">SGI implementation</a>, and
 401   ported to many other platforms. It includes a debug mode that uses a
 402   wrapper model (that in some ways inspired the libstdc++ debug mode
 403   design), although at the time of this writing the debug mode is
 404   somewhat incomplete and meets only the "Full user recompilation" (2)
 405   recompilation guarantee by requiring the user to link against a
 406   different library in debug mode vs. release mode.</p></li><li class="listitem"><p>Metrowerks CodeWarrior: The C++ standard library
 407   that ships with Metrowerks CodeWarrior includes a debug mode. It is
 408   a full debug-mode implementation (including debugging for
 409   CodeWarrior extensions) and is easy to use, although it meets only
 410   the "Full recompilation" (1) recompilation
 411   guarantee.</p></li></ul></div></div></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="debug_mode_using.html">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="debug_mode.html">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="parallel_mode.html">Next</a></td></tr><tr><td width="40%" align="left" valign="top">Using </td><td width="20%" align="center"><a accesskey="h" href="../index.html">Home</a></td><td width="40%" align="right" valign="top"> Chapter 18. Parallel Mode</td></tr></table></div></body></html>