Re: [boost] Re: [serialization] and macro 'max' compiler warning on VC71

From: Cory Nelson (phrosty_at_[hidden])
Date: 2004-11-22 16:38:53

This is a problem with windows.h, it includes the min() and max()
macros which can easily screw over c++.  Define NOMINMAX before
windows.h is included and the problem goes away.

On Mon, 22 Nov 2004 14:56:26 -0500, Jeff Flinn
<triumphsprint2000_at_[hidden]> wrote:
> 
> "Robert Ramey" <ramey_at_[hidden]> wrote in message
> news:cntdbc$bah$1_at_sea.gmane.org...
> > I use MFC on a regular basis.  What works best for me is to not mix boost
> > and MFC headers in the same module.  This requires a little bit of
> > organization but I believe is simpler in the long run.
> 
> Except when one is serializing views themselves.
> 
> 
> 
> > Robert Ramey
> >
> > "Jeff Flinn" <TriumphSprint2000_at_[hidden]> wrote in message
> > news:cntc3l$7ia$1_at_sea.gmane.org...
> > I've just tried to move to 1.32 and get the following warning:
> >
> > C:\boost\boost_1_32_0\boost\archive\basic_binary_oprimitive.hpp(101) :
> > warning C4003: not enough actual parameters for macro 'max'
> >
> > which points to:
> >
> >     assert(
> >
> > static_cast<std::size_t>(std::numeric_limits<std::streamsize>::max()) >=
> > count
> >     );
> 
> I seemed to remember related min/max discussions in the past that suggested
> surrounding the qualified function name in parens as shown below
> 
> assert
> (
> static_cast<std::size_t>((std::numeric_limits<std::streamsize>::max)()) >=
> count
> );
> 
> This appears to work in both binary_oprimitive line 101 and
> binary_iprimitive line 105
> 
> 
> 
> Thanks, Jeff
> 
> _______________________________________________
> Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
> 

-- 
Cory Nelson
http://www.int64.org