Re: Build failure on Windows using MSVC

[dhruva]

Hi Eli,
 This is the error message I get when I get to compile. I agree WINVER is 
deprecated but _WIN32_WINNT is not defined after including windows..h, I 
therefore used WINVER instead which is defined. For sake of completeness, the 
windows.h comes from the PlatformSDK installed as part of Visual Studio 
installation.
 Another alternative at fixing this would be to test it in configure. That may 
be a foolproof approach.

The compilation error I notice:

        cl -I. -nologo -c -Zel -W2 -H63 -Oxsb2 -Oy- -G6dF -Zp8 -Zi   
/Ic:/tools/msvcimage/include -Demacs=1 -DHAVE_CONFIG_H -I../nt/inc 
-DHAVE_NTGUI=1  -Foobj-spd/i386\ w32.c
w32.c
w32.c(78) : error C2011: '_MEMORYSTATUSEX' : 'struct' type redefinition
        C:\Program Files\Microsoft Visual Studio .NET 
2003\Vc7\PlatformSDK\Include\WinBase.h(1723) : see declaration of 
'_MEMORYSTATUSEX'
w32.c(3605) : warning C4244: '=' : conversion from 'long double' to 'long', 
possible loss of data
w32.c(3606) : warning C4244: '=' : conversion from 'long double' to 'long', 
possible loss of data
w32.c(3609) : warning C4244: '=' : conversion from 'long double' to 'long', 
possible loss of data
w32.c(3610) : warning C4244: '=' : conversion from 'long double' to 'long', 
possible loss of data
w32.c(3616) : warning C4244: '=' : conversion from 'long double' to 'long', 
possible loss of data
w32.c(3617) : warning C4244: '=' : conversion from 'long double' to 'long', 
possible loss of data
w32.c(3621) : warning C4244: '=' : conversion from 'long double' to 'long', 
possible loss of data
w32.c(3622) : warning C4244: '=' : conversion from 'long double' to 'long', 
possible loss of data
w32.c(3668) : warning C4244: '=' : conversion from 'double' to 'DWORD', 
possible loss of data
NMAKE : fatal error U1077: 'cl' : return code '0x2'
Stop.
NMAKE : fatal error U1077: '"C:\Program Files\Microsoft Visual Studio .NET 
2003\VC7\BIN\nmake.exe"' : return code '0x2'
Stop.

-dhruva



----- Original Message ----
> From: Eli Zaretskii <address@hidden>
> To: dhruva <address@hidden>
> Cc: address@hidden
> Sent: Friday, 15 August, 2008 3:01:13 PM
> Subject: Re: Build failure on Windows using MSVC
> 
> > Date: Fri, 15 Aug 2008 13:19:37 +0530 (IST)
> > From: dhruva 
> > 
> >  Build of Emacs from HEAD on WXP using MSVC 2003 fails due to redefinition 
> > of 
> a structure MEMORYSTATUSEX. The following patch fixes the issue on WXP (W2K 
> prof 
> and upwards as that structure is available on W2K+). Need to test on older 
> versions and I have no access to it.
> 
> What is the exact error message emitted by Studio 2003?  Is there
> perhaps a compiler switch to avoid such an error?  Is our definition
> incompatible with what VS 2003 defines?
> 
> I need the above information to come up with a suitable solution that
> will work with all supported platforms and compiler versions.  Someone
> else compiled this CVS code with VS 6 and didn't have this problem, so
> we need to find a solution that doesn't break versions of VS other
> than yours.
> 
> > === modified file 'src/w32.c'
> > --- src/w32.c   2008-08-10 01:05:02 +0000
> > +++ src/w32..c   2008-08-15 07:43:51 +0000
> > @@ -73,6 +73,7 @@
> >  #define _ANONYMOUS_STRUCT
> >  #endif
> >  #include 
> > +#if (WINVER<0x0500)
> 
> I don't want to use WINVER.  It's an old and deprecated macro, and we
> already use its more modern replacement _WIN32_WINNT.  Also, the
> availability of the function at run time is not an issue here, since
> the code dynamically tests that and only uses the function if it's
> available.  The issue is compilation.
> 
> From your report, I understand that the MS header files do not guard
> the declaration of MEMORYSTATUSEX with _WIN32_WINNT, like the MinGW
> headers do?  Can you verify that, please?
> 
> If this is so, I prefer to condition the declaration on __MINGW32__
> rather than use WINVER.  The only thing that bothers me is VS 6 (see
> above).



      Unlimited freedom, unlimited storage. Get it now, on 
http://help.yahoo.com/l/in/yahoo/mail/yahoomail/tools/tools-08.html/