Re: [Boost-testing] [Sandia-Sun][RudbekAssociates-V2] Spurious config test results on some platforms?

From: John Maddock (john_at_[hidden])
Date: 2007-10-01 10:46:25

Victor A. Wagner Jr. wrote:
> for version 7.1: Microsoft (R) 32-bit C/C++ Optimizing Compiler
> Version 14.00.50727.762 for 80x86

That's a VC8 version, but your config_info results are reporting version 
1310 so I assume that's a typo on your part :-)

> for version 8.0:  Microsoft (R) 32-bit C/C++ Optimizing Compiler
> Version 14.00.50727.762 for 80x86

That's the same version number I have here.

> from visualc.hpp:
> #if _MSC_VER <= 1400  // 1400 == VC++ 8.0
> #  define BOOST_NO_MEMBER_TEMPLATE_FRIENDS
> #  define BOOST_NO_TWO_PHASE_NAME_LOOKUP
> #endif
>
>
> so....It's defined for one, and not the other

No: note <= 1400 so its defined for *both*, but none the less *both* are 
failing, and neither is reporting BOOST_NO_TWO_PHASE_NAME_LOOKUP as defined 
in their config_info results.

Any ideas?  Are the tests run incrementally?  It shouldn't actually make any 
difference since I "touched" config_test.cpp to force test runners to 
rebuild it (bjam can't trace dependencies through the config headers).

Thanks, John.