Boost_LOG_DOXYGEN_PASS在Boost.Log中是什么？

Question

目前正处在Boost.Log库中一些奇怪的代码，例如在BoostFileLogging.cpp，它说

#ifndef BOOST_LOG_DOXYGEN_PASS 

#define BOOST_LOG_INIT_LOG_TO_FILE_INTERNAL(z, n, data)\ 
    template< BOOST_PP_ENUM_PARAMS(n, typename T) >\ 
    inline shared_ptr< BOOST_LOG_FILE_SINK_FRONTEND_INTERNAL<sinks::text_file_backend> > add_file_log(BOOST_PP_ENUM_BINARY_PARAMS(n, T, const& arg))\ 
    {\ 
     return aux::add_file_log((\ 
      aux::wrap_file_name(arg0, typename parameter::aux::is_named_argument<T0>::type())\ 
      BOOST_PP_COMMA_IF(BOOST_PP_GREATER(n, 1))\ 
      BOOST_PP_ENUM_SHIFTED_PARAMS(n, arg)\ 
      ));\ 
    } 

BOOST_PP_REPEAT_FROM_TO(1, BOOST_LOG_MAX_PARAMETER_ARGS, BOOST_LOG_INIT_LOG_TO_FILE_INTERNAL, ~) 

#undef BOOST_LOG_INIT_LOG_TO_FILE_INTERNAL 

#else // BOOST_LOG_DOXYGEN_PASS 

/*! 
* The function initializes the logging library to write logs to a file stream. 
* 
* \param args A number of named arguments. The following parameters are supported: 
*    \li \c file_name The file name or its pattern. This parameter is mandatory. 
*    \li \c open_mode The mask that describes the open mode for the file. See <tt>std::ios_base::openmode</tt>. 
*    \li \c rotation_size The size of the file at which rotation should occur. See <tt>basic_text_file_backend</tt>. 
*    \li \c time_based_rotation The predicate for time-based file rotations. See <tt>basic_text_file_backend</tt>. 
*    \li \c auto_flush A boolean flag that shows whether the sink should automatically flush the file 
*        after each written record. 
*    \li \c target The target directory to store rotated files in. See <tt>sinks::file::make_collector</tt>. 
*    \li \c max_size The maximum total size of rotated files in the target directory. See <tt>sinks::file::make_collector</tt>. 
*    \li \c min_free_space Minimum free space in the target directory. See <tt>sinks::file::make_collector</tt>. 
*    \li \c scan_method The method of scanning the target directory for log files. See <tt>sinks::file::scan_method</tt>. 
*    \li \c filter Specifies a filter to install into the sink. May be a string that represents a filter, 
*       or a filter lambda expression. 
*    \li \c format Specifies a formatter to install into the sink. May be a string that represents a formatter, 
*       or a formatter lambda expression (either streaming or Boost.Format-like notation). 
* \return Pointer to the constructed sink. 
*/ 
template< typename... ArgsT > 
shared_ptr< BOOST_LOG_FILE_SINK_FRONTEND_INTERNAL<sinks::text_file_backend> > add_file_log(ArgsT... const& args); 

#endif // BOOST_LOG_DOXYGEN_PASS 

什么是该宏的动机是什么？我应该启用它还是禁用它？它没有记录。

Answer 1

Doxygen是一个解析源代码并生成文档的程序。

这样做的目的是让Doxygen看到（和记录）与真实编译器看到的东西不同的东西。

在这种情况下，使用Doxygen会看到一个可变参数模板，因此HTML文档会说这样的事情：

add_file_log（ArgsT ...常量&参数）;

模板参数：

• 模板<类型名称... ArgsT>

返回类型：

shared_ptr< BOOST_LOG_FILE_SINK_FRONTEND_INTERNAL<sinks::text_file_backend> > 

函数初始化日志库日志写入文件流。

...等等

但升压登录显然适用于不支持可变参数模板编译器，所以真正的编译器看不到这一点。什么编译器看到的是一堆非可变参数模板，东西沿着这些线路（虽然它是不相关的解释BOOST_LOG_DOXYGEN_PASS）：

template<typename Arg1T> shared_ptr</*snip*/> add_file_log(Arg1T const& arg1); 
template<typename Arg1T, typename Arg2T> shared_ptr</*snip*/> add_file_log(Arg1T const& arg1, Arg2T const& arg2); 
template<typename Arg1T, typename Arg2T, typename Arg3T> shared_ptr</*snip*/> add_file_log(Arg1T const& arg1, Arg2T const& arg2, Arg3T const& arg3); 
// and so on 

你不会想对于这些独立的文档记录，那么它将只是杂乱的文件。