John Maddock wrote:

Following on from the recent discussion, we now have:

BOOST_NO_IS_ABSTRACT added to the config system, this is likely to cause some new failures with config_test.cpp until we get the macro added to those compiler config files that need it (although Gcc, Visual C++, Borland C++ are already taken care of).

is_abstract no longer uses SFINAE when BOOST_NO_IS_ABSTRACT is defined, instead it defaults to the same value as is_polymorphic: the rationale is that "true" is the safe default (if there is one), but since non-polymorphic types obviously can't be abstract,

struct X { protected: X(); ~X(); }; Not abstract? I guess the trait can't detect that one anyway, though. -- Dave Abrahams Boost Consulting http://www.boost-consulting.com

John Maddock wrote:

Following on from the recent discussion, we now have:

BOOST_NO_IS_ABSTRACT added to the config system, this is likely to cause some new failures with config_test.cpp until we get the macro added to those compiler config files that need it (although Gcc, Visual C++, Borland C++ are already taken care of).

I didn't follow this discussion ... please could somebody write me two sentences about how to make cw* pass again? Stefan