Re: [boost] [1.35][lexical_cast][concept check][range] Build errorandwarnings with MSVC8 on Warning Level 4

From: Stephan Diederich (stephan.diederich_at_[hidden])
Date: 2008-04-09 20:05:00

On Wed, Apr 9, 2008 at 6:36 PM, Paul A Bristow <pbristow_at_[hidden]> wrote:
>
>
>  >-----Original Message-----
>  >From: boost-bounces_at_[hidden]
>  >[mailto:boost-bounces_at_[hidden]] On Behalf Of Stephan Diederich
>
> >Sent: 09 April 2008 17:24
>  >To: boost_at_[hidden]
>
>
> >Subject: Re: [boost] [1.35][lexical_cast][concept
>  >check][range] Build errorandwarnings with MSVC8 on Warning Level 4
>  >
>  >> >As we tried to update to 1.35 today, we found some warnings
>  >with MSVC
>  >> >on Warning Level 4 in:
>  >> >
>  >> >lexical_cast:
>  >> >There was a return statement after a call to throw_exception which
>  >> >causes a unreachable code warning.
>  >> >I can see no problem by just removing that return.
>  >> >see  http://svn.boost.org/trac/boost/ticket/1791
>  >>
>  >> The problem code is:
>  >>
>  >> throw_exception(bad_lexical_cast(typeid(Source), typeid(Target)));
>  >> return Target(); // normally never reached (throw_exception)
>  >>
>  >> Is the problem that the compiler can't be confident that
>  >throw_exception won't ever return?  (For that matter, can we?)
>  >
>  >Hm, I'm not sure what you mean. Do you think we exchange one warning
>  >with another? "Code not reachable" with "Missing return value"?.
>  >I think _we_ can be sure throw_exception does not return.
>  >throw_exception doc ( http://tinyurl.com/4vup33 ) says that even the
>  >user defined one is not allowed to return.
>
>  Well it actually says
>
>  "Callers of throw_exception  are allowed to assume that the function never returns; therefore, if the user-defined throw_exception
>  returns, the behavior is undefined."
>
>  'Undefined' is not quite the same as 'won't happen'.
>
>  In the worst case, the result of excuting the return statement might be less worse ;-)
>
>  So is it still safer to keep the return?  One statement is quite cheap ;-)

I totally agree.

Would be nice to see the warning removed.

cheers,
stephan

>  >> with a push'n'pop around this:
>  >>
>  >> #if (defined _MSC_VER)
>  >> # pragma warning( push )
>  >> # pragma warning( disable : 4702 ) : unreachable code
>  >> #endif
>  >>
>  >> throw_exception(bad_lexical_cast(typeid(Source), typeid(Target)));
>  >> return Target(); // normally never reached (throw_exception)
>  >>
>  >> #if (defined _MSC_VER)
>  >> # pragma warning( pop )
>  >> #endif