#ifndef RBIMPL_ATTR_NOEXCEPT_H /*-*-C++-*-vi:se ft=cpp:*/ #define RBIMPL_ATTR_NOEXCEPT_H /** * @file * @author Ruby developers * @copyright This file is a part of the programming language Ruby. * Permission is hereby granted, to either redistribute and/or * modify this file, provided that the conditions mentioned in the * file COPYING are met. Consult the file for details. * @warning Symbols prefixed with either `RBIMPL` or `rbimpl` are * implementation details. Don't take them as canon. They could * rapidly appear then vanish. The name (path) of this header file * is also an implementation detail. Do not expect it to persist * at the place it is now. Developers are free to move it anywhere * anytime at will. * @note To ruby-core: remember that this header can be possibly * recursively included from extension libraries written in C++. * Do not expect for instance `__VA_ARGS__` is always available. * We assume C99 for ruby itself but we don't assume languages of * extension libraries. They could be written in C++98. * @brief Defines #RBIMPL_ATTR_NOEXCEPT. * * This isn't actually an attribute in C++ but who cares... * * Mainly due to aesthetic reasons, this one is rarely used in the project. * But can be handy on occasions, especially when a function's noexcept-ness * depends on its calling functions. * * ### Q&A ### * * - Q: Can a function that raises Ruby exceptions be attributed `noexcept`? * * - A: Yes. `noexcept` is about C++ exceptions, not Ruby's. They don't * interface each other. You can safely attribute a function that raises * Ruby exceptions as `noexcept`. * * - Q: How, then, can I assert that a function I wrote doesn't raise any Ruby * exceptions? * * - A: `__attribute__((__leaf__))` is for that purpose. A function attributed * as leaf can still throw C++ exceptions, but not Ruby's. Note however, * that it's extremely difficult -- if not impossible -- to assert that a * function doesn't raise any Ruby exceptions at all. Use of that * attribute is not recommended; mere mortals can't properly use that by * hand. * * - Q: Does it make sense to attribute an inline function `noexcept`? * * - A: I thought so before. But no, I don't think they are useful any longer. * * - When an inline function attributed `noexcept` actually doesn't throw * any exceptions at all: these days I don't see any difference in * generated assembly by adding/removing this attribute. C++ compilers * get smarter and smarter. Today they can infer if it actually throws * or not without any annotations by humans (correct me if I'm wrong). * * - When an inline function attributed `noexcepr` actually _does_ throw an * exception: they have to call `std::terminate` then (C++ standard * mandates so). This means exception handling routines are actually * enforced, not omitted. This doesn't impact runtime performance (The * Itanium C++ ABI has zero-cost exception handling), but does impact on * generated binary size. This is bad. */ #include "ruby/internal/compiler_since.h" #include "ruby/internal/has/feature.h" /** Wraps (or simulates) C++11 `noexcept` */ #if ! defined(__cplusplus) # /* Doesn't make sense. */ # define RBIMPL_ATTR_NOEXCEPT(_) /* void */ #elif RBIMPL_HAS_FEATURE(cxx_noexcept) # define RBIMPL_ATTR_NOEXCEPT(_) noexcept(noexcept(_)) #elif defined(__GXX_EXPERIMENTAL_CXX0X__) && __GXX_EXPERIMENTAL_CXX0X__ # define RBIMPL_ATTR_NOEXCEPT(_) noexcept(noexcept(_)) #elif defined(__INTEL_CXX11_MODE__) # define RBIMPL_ATTR_NOEXCEPT(_) noexcept(noexcept(_)) #elif RBIMPL_COMPILER_SINCE(MSVC, 19, 0, 0) # define RBIMPL_ATTR_NOEXCEPT(_) noexcept(noexcept(_)) #elif __cplusplus >= 201103L # define RBIMPL_ATTR_NOEXCEPT(_) noexcept(noexcept(_)) #else # define RBIMPL_ATTR_NOEXCEPT(_) /* void */ #endif #endif /* RBIMPL_ATTR_NOEXCEPT_H */