Douglas Pratley wrote: > The logic of this looks wrong to me in one sense - the message tells you > that your compiler is 14.00 when it is not (I'm still using 12.00). I > would have thought it should be reversed, so that given a compiler > number, it checks that the variant is one of the accepted values. > > More importantly, I tried adding these lines: > > #define WS_TO_STRING2(x) #x > #define WS_TO_STRING(x) WS_TO_STRING2(x) > #pragma message( "_MSC_VER: " WS_TO_STRING(_MSC_VER)) > #pragma message( "MSVC_VARIANT: " WS_TO_STRING(MSVC_VARIANT) ) > > I then get: > > _MSC_VER: 1200 > MSVC_VARIANT: MSVC_VARIANT > > I don't think that MSVC_VARIANT, or any of the constants it's being > compared against, are actually getting defined. > > I can't see anything in config.nmake that would actually cause them to > be passed to the preprocessor. Am I missing something? > Could you check 20614 instead? At least I've fixed the check logic in config.h here.
As you point it out, I don't know why the config.h lines: #elseif _MSC_VER == 1200 && (MSVC_VARIANT != MSVC6) #error Your MSVC_VARIANT setting in config.nmake and the C compiler version 12.00 (V6.0) don't match! don't trigger an error now (it doesn't do in the builtbot) - as MSVC_VARIANT isn't defined at all, it shouldn't be MSVC6 ?!? We might need to add something like: -DMSVC_VARIANT=$(MSVC_VARIANT) (I don't have the right syntax at hand) to the Makefile.nmake CFLAGS. I currently don't really know why it currently doesn't throw a compiler error - but it obviously don't do it now (seems I'm missing something here?!?) ... Regards, ULFL _______________________________________________ Wireshark-dev mailing list Wireshark-dev@wireshark.org http://www.wireshark.org/mailman/listinfo/wireshark-dev
