Matthias Troyer <[EMAIL PROTECTED]> writes:
The third patch (patch3) is a workaround for operators.hpp, since the Boost PP library does not seem to work on the IBM. Is the preprocessor library really needed for that name mangling:
#define BOOST_OPERATOR2_LEFT(name) BOOST_PP_SEQ_CAT_S(1,(name)(2)(_)(left))
or couldn't the simple
#define BOOST_OPERATOR2_LEFT(name) name##2_left
be used which would make Boost more portable? If there are reasons why the PP version is preferred? If there is a reason for the PP version then could someone please apply the patch with the workaround?
Done.
Looks good to me, too.
Daniel, I also had to apply massive patching to accessibility in bool_testable. It fails on so many compilers that I begin to have doubts that making it private is legal. Could you check, please?
It seems to me that replacing the plain "operator bool()" by the safe-bool-idiom in Wrapper1 caused the problems. This can't work, as bool_testable tries to provide safe bool conversion, too, and both idioms together are incompatible. In fact, bool_testable was meant to replace the safe-bool-idiom, making it easier for users to write their operator bool(). Maybe I should clarify this in the documentation to prevent others from making the same error. Can you please correct it and see if the problems you were seeing remain?
Regards, Daniel
-- Daniel Frey
aixigo AG - financial training, research and technology Schloß-Rahe-Straße 15, 52072 Aachen, Germany fon: +49 (0)241 936737-42, fax: +49 (0)241 936737-99 eMail: [EMAIL PROTECTED], web: http://www.aixigo.de
_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
