Mark throw_exception as noreturn?
Hi, I was wondering why the throw_exception function is not marked as noreturn? This makes it troublesome to use it as a drop-in replacement for throw statements as it starts to flag compiler warnings about missing return statements in non-void functions. I suggest to mark the function as noreturn at least for compilers that support it. GCC has __attribute__((noreturn)), MSVC has __declspec(noreturn).
Andrey Semashev wrote:
Hi,
I was wondering why the throw_exception function is not marked as noreturn?
I don't know. That could be because not all compilers support that annotation (but then, that's not usually something which scares Boost off).
This makes it troublesome to use it as a drop-in replacement for throw statements as it starts to flag compiler warnings about missing return statements in non-void functions.
I suggest to mark the function as noreturn at least for compilers that support it. GCC has __attribute__((noreturn)), MSVC has __declspec(noreturn).
I don't know what they guys here want to do, or whether there's an accepted solution to this. For the immediate needs, however, you could use BOOST_UNREACHABLE_RETURN in the calling function. -- Genny
Gennaro Prota wrote:
For the immediate needs, however, you could use BOOST_UNREACHABLE_RETURN in the calling function.
From what I see from Boost.Config headers, the macro unfolds to something non-empty only for three compilers, and GCC is not among them. I encountered the problem with GCC.
Andrey Semashev wrote:
I was wondering why the throw_exception function is not marked as noreturn? This makes it troublesome to use it as a drop-in replacement for throw statements as it starts to flag compiler warnings about missing return statements in non-void functions.
I suggest to mark the function as noreturn at least for compilers that support it. GCC has __attribute__((noreturn)), MSVC has __declspec(noreturn).
Why is throw_exception detected as returning something in the first place? I've personally tried to force __attribute__((noreturn)) with GCC, but it doesn't work, it says the function does return. Looking at the code though, I can't tell what is wrong with it. A fix to that would be appreciated. Without that, I'd personally rather not use throw_exception.
On Sat, Jul 4, 2009 at 6:11 PM, Mathias Gaunard<mathias.gaunard@ens-lyon.org> wrote:
Andrey Semashev wrote:
I was wondering why the throw_exception function is not marked as noreturn? This makes it troublesome to use it as a drop-in replacement for throw statements as it starts to flag compiler warnings about missing return statements in non-void functions.
I suggest to mark the function as noreturn at least for compilers that support it. GCC has __attribute__((noreturn)), MSVC has __declspec(noreturn).
Why is throw_exception detected as returning something in the first place?
I've personally tried to force __attribute__((noreturn)) with GCC, but it doesn't work, it says the function does return.
Looking at the code though, I can't tell what is wrong with it. A fix to that would be appreciated. Without that, I'd personally rather not use throw_exception.
I've added a macro BOOST_ATTRIBUTE_NORETURN. The problem is fixed for MSVC, feel free to add support for other compilers in boost/exception/detail/attribute_noreturn.hpp. Emil Dotchevski Reverge Studios, Inc. http://www.revergestudios.com/reblog/index.php?n=ReCode
Emil Dotchevski wrote:
I've added a macro BOOST_ATTRIBUTE_NORETURN. The problem is fixed for MSVC, feel free to add support for other compilers in boost/exception/detail/attribute_noreturn.hpp.
Should it really be part of Boost.Exception (moreover, private part)? Maybe it's better to be in Boost.Config?
On Sun, Jul 5, 2009 at 2:13 AM, Andrey Semashev<andrey.semashev@gmail.com> wrote:
Emil Dotchevski wrote:
I've added a macro BOOST_ATTRIBUTE_NORETURN. The problem is fixed for MSVC, feel free to add support for other compilers in boost/exception/detail/attribute_noreturn.hpp.
Should it really be part of Boost.Exception (moreover, private part)? Maybe it's better to be in Boost.Config?
Probably the correct place is Boost.Config, but so far boost::throw_exception (which is part of Boost Exception) seems to be the only function that needs this macro and I thought we don't have good enough reason to add weight to Boost.Config at this point. It'll take no time at all to make BOOST_ATTRIBUTE_NORETURN official if it is needed elsewhere in the future. Emil Dotchevski Reverge Studios, Inc. http://www.revergestudios.com/reblog/index.php?n=ReCode
participants (4)
-
Andrey Semashev -
Emil Dotchevski -
Gennaro Prota -
Mathias Gaunard